summaryrefslogtreecommitdiff
path: root/systems/knuth/dist/tex/texbook.tex
blob: a1e50786ead173af9523d90b60a4806f67ddef4b (plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
1001
1002
1003
1004
1005
1006
1007
1008
1009
1010
1011
1012
1013
1014
1015
1016
1017
1018
1019
1020
1021
1022
1023
1024
1025
1026
1027
1028
1029
1030
1031
1032
1033
1034
1035
1036
1037
1038
1039
1040
1041
1042
1043
1044
1045
1046
1047
1048
1049
1050
1051
1052
1053
1054
1055
1056
1057
1058
1059
1060
1061
1062
1063
1064
1065
1066
1067
1068
1069
1070
1071
1072
1073
1074
1075
1076
1077
1078
1079
1080
1081
1082
1083
1084
1085
1086
1087
1088
1089
1090
1091
1092
1093
1094
1095
1096
1097
1098
1099
1100
1101
1102
1103
1104
1105
1106
1107
1108
1109
1110
1111
1112
1113
1114
1115
1116
1117
1118
1119
1120
1121
1122
1123
1124
1125
1126
1127
1128
1129
1130
1131
1132
1133
1134
1135
1136
1137
1138
1139
1140
1141
1142
1143
1144
1145
1146
1147
1148
1149
1150
1151
1152
1153
1154
1155
1156
1157
1158
1159
1160
1161
1162
1163
1164
1165
1166
1167
1168
1169
1170
1171
1172
1173
1174
1175
1176
1177
1178
1179
1180
1181
1182
1183
1184
1185
1186
1187
1188
1189
1190
1191
1192
1193
1194
1195
1196
1197
1198
1199
1200
1201
1202
1203
1204
1205
1206
1207
1208
1209
1210
1211
1212
1213
1214
1215
1216
1217
1218
1219
1220
1221
1222
1223
1224
1225
1226
1227
1228
1229
1230
1231
1232
1233
1234
1235
1236
1237
1238
1239
1240
1241
1242
1243
1244
1245
1246
1247
1248
1249
1250
1251
1252
1253
1254
1255
1256
1257
1258
1259
1260
1261
1262
1263
1264
1265
1266
1267
1268
1269
1270
1271
1272
1273
1274
1275
1276
1277
1278
1279
1280
1281
1282
1283
1284
1285
1286
1287
1288
1289
1290
1291
1292
1293
1294
1295
1296
1297
1298
1299
1300
1301
1302
1303
1304
1305
1306
1307
1308
1309
1310
1311
1312
1313
1314
1315
1316
1317
1318
1319
1320
1321
1322
1323
1324
1325
1326
1327
1328
1329
1330
1331
1332
1333
1334
1335
1336
1337
1338
1339
1340
1341
1342
1343
1344
1345
1346
1347
1348
1349
1350
1351
1352
1353
1354
1355
1356
1357
1358
1359
1360
1361
1362
1363
1364
1365
1366
1367
1368
1369
1370
1371
1372
1373
1374
1375
1376
1377
1378
1379
1380
1381
1382
1383
1384
1385
1386
1387
1388
1389
1390
1391
1392
1393
1394
1395
1396
1397
1398
1399
1400
1401
1402
1403
1404
1405
1406
1407
1408
1409
1410
1411
1412
1413
1414
1415
1416
1417
1418
1419
1420
1421
1422
1423
1424
1425
1426
1427
1428
1429
1430
1431
1432
1433
1434
1435
1436
1437
1438
1439
1440
1441
1442
1443
1444
1445
1446
1447
1448
1449
1450
1451
1452
1453
1454
1455
1456
1457
1458
1459
1460
1461
1462
1463
1464
1465
1466
1467
1468
1469
1470
1471
1472
1473
1474
1475
1476
1477
1478
1479
1480
1481
1482
1483
1484
1485
1486
1487
1488
1489
1490
1491
1492
1493
1494
1495
1496
1497
1498
1499
1500
1501
1502
1503
1504
1505
1506
1507
1508
1509
1510
1511
1512
1513
1514
1515
1516
1517
1518
1519
1520
1521
1522
1523
1524
1525
1526
1527
1528
1529
1530
1531
1532
1533
1534
1535
1536
1537
1538
1539
1540
1541
1542
1543
1544
1545
1546
1547
1548
1549
1550
1551
1552
1553
1554
1555
1556
1557
1558
1559
1560
1561
1562
1563
1564
1565
1566
1567
1568
1569
1570
1571
1572
1573
1574
1575
1576
1577
1578
1579
1580
1581
1582
1583
1584
1585
1586
1587
1588
1589
1590
1591
1592
1593
1594
1595
1596
1597
1598
1599
1600
1601
1602
1603
1604
1605
1606
1607
1608
1609
1610
1611
1612
1613
1614
1615
1616
1617
1618
1619
1620
1621
1622
1623
1624
1625
1626
1627
1628
1629
1630
1631
1632
1633
1634
1635
1636
1637
1638
1639
1640
1641
1642
1643
1644
1645
1646
1647
1648
1649
1650
1651
1652
1653
1654
1655
1656
1657
1658
1659
1660
1661
1662
1663
1664
1665
1666
1667
1668
1669
1670
1671
1672
1673
1674
1675
1676
1677
1678
1679
1680
1681
1682
1683
1684
1685
1686
1687
1688
1689
1690
1691
1692
1693
1694
1695
1696
1697
1698
1699
1700
1701
1702
1703
1704
1705
1706
1707
1708
1709
1710
1711
1712
1713
1714
1715
1716
1717
1718
1719
1720
1721
1722
1723
1724
1725
1726
1727
1728
1729
1730
1731
1732
1733
1734
1735
1736
1737
1738
1739
1740
1741
1742
1743
1744
1745
1746
1747
1748
1749
1750
1751
1752
1753
1754
1755
1756
1757
1758
1759
1760
1761
1762
1763
1764
1765
1766
1767
1768
1769
1770
1771
1772
1773
1774
1775
1776
1777
1778
1779
1780
1781
1782
1783
1784
1785
1786
1787
1788
1789
1790
1791
1792
1793
1794
1795
1796
1797
1798
1799
1800
1801
1802
1803
1804
1805
1806
1807
1808
1809
1810
1811
1812
1813
1814
1815
1816
1817
1818
1819
1820
1821
1822
1823
1824
1825
1826
1827
1828
1829
1830
1831
1832
1833
1834
1835
1836
1837
1838
1839
1840
1841
1842
1843
1844
1845
1846
1847
1848
1849
1850
1851
1852
1853
1854
1855
1856
1857
1858
1859
1860
1861
1862
1863
1864
1865
1866
1867
1868
1869
1870
1871
1872
1873
1874
1875
1876
1877
1878
1879
1880
1881
1882
1883
1884
1885
1886
1887
1888
1889
1890
1891
1892
1893
1894
1895
1896
1897
1898
1899
1900
1901
1902
1903
1904
1905
1906
1907
1908
1909
1910
1911
1912
1913
1914
1915
1916
1917
1918
1919
1920
1921
1922
1923
1924
1925
1926
1927
1928
1929
1930
1931
1932
1933
1934
1935
1936
1937
1938
1939
1940
1941
1942
1943
1944
1945
1946
1947
1948
1949
1950
1951
1952
1953
1954
1955
1956
1957
1958
1959
1960
1961
1962
1963
1964
1965
1966
1967
1968
1969
1970
1971
1972
1973
1974
1975
1976
1977
1978
1979
1980
1981
1982
1983
1984
1985
1986
1987
1988
1989
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
2015
2016
2017
2018
2019
2020
2021
2022
2023
2024
2025
2026
2027
2028
2029
2030
2031
2032
2033
2034
2035
2036
2037
2038
2039
2040
2041
2042
2043
2044
2045
2046
2047
2048
2049
2050
2051
2052
2053
2054
2055
2056
2057
2058
2059
2060
2061
2062
2063
2064
2065
2066
2067
2068
2069
2070
2071
2072
2073
2074
2075
2076
2077
2078
2079
2080
2081
2082
2083
2084
2085
2086
2087
2088
2089
2090
2091
2092
2093
2094
2095
2096
2097
2098
2099
2100
2101
2102
2103
2104
2105
2106
2107
2108
2109
2110
2111
2112
2113
2114
2115
2116
2117
2118
2119
2120
2121
2122
2123
2124
2125
2126
2127
2128
2129
2130
2131
2132
2133
2134
2135
2136
2137
2138
2139
2140
2141
2142
2143
2144
2145
2146
2147
2148
2149
2150
2151
2152
2153
2154
2155
2156
2157
2158
2159
2160
2161
2162
2163
2164
2165
2166
2167
2168
2169
2170
2171
2172
2173
2174
2175
2176
2177
2178
2179
2180
2181
2182
2183
2184
2185
2186
2187
2188
2189
2190
2191
2192
2193
2194
2195
2196
2197
2198
2199
2200
2201
2202
2203
2204
2205
2206
2207
2208
2209
2210
2211
2212
2213
2214
2215
2216
2217
2218
2219
2220
2221
2222
2223
2224
2225
2226
2227
2228
2229
2230
2231
2232
2233
2234
2235
2236
2237
2238
2239
2240
2241
2242
2243
2244
2245
2246
2247
2248
2249
2250
2251
2252
2253
2254
2255
2256
2257
2258
2259
2260
2261
2262
2263
2264
2265
2266
2267
2268
2269
2270
2271
2272
2273
2274
2275
2276
2277
2278
2279
2280
2281
2282
2283
2284
2285
2286
2287
2288
2289
2290
2291
2292
2293
2294
2295
2296
2297
2298
2299
2300
2301
2302
2303
2304
2305
2306
2307
2308
2309
2310
2311
2312
2313
2314
2315
2316
2317
2318
2319
2320
2321
2322
2323
2324
2325
2326
2327
2328
2329
2330
2331
2332
2333
2334
2335
2336
2337
2338
2339
2340
2341
2342
2343
2344
2345
2346
2347
2348
2349
2350
2351
2352
2353
2354
2355
2356
2357
2358
2359
2360
2361
2362
2363
2364
2365
2366
2367
2368
2369
2370
2371
2372
2373
2374
2375
2376
2377
2378
2379
2380
2381
2382
2383
2384
2385
2386
2387
2388
2389
2390
2391
2392
2393
2394
2395
2396
2397
2398
2399
2400
2401
2402
2403
2404
2405
2406
2407
2408
2409
2410
2411
2412
2413
2414
2415
2416
2417
2418
2419
2420
2421
2422
2423
2424
2425
2426
2427
2428
2429
2430
2431
2432
2433
2434
2435
2436
2437
2438
2439
2440
2441
2442
2443
2444
2445
2446
2447
2448
2449
2450
2451
2452
2453
2454
2455
2456
2457
2458
2459
2460
2461
2462
2463
2464
2465
2466
2467
2468
2469
2470
2471
2472
2473
2474
2475
2476
2477
2478
2479
2480
2481
2482
2483
2484
2485
2486
2487
2488
2489
2490
2491
2492
2493
2494
2495
2496
2497
2498
2499
2500
2501
2502
2503
2504
2505
2506
2507
2508
2509
2510
2511
2512
2513
2514
2515
2516
2517
2518
2519
2520
2521
2522
2523
2524
2525
2526
2527
2528
2529
2530
2531
2532
2533
2534
2535
2536
2537
2538
2539
2540
2541
2542
2543
2544
2545
2546
2547
2548
2549
2550
2551
2552
2553
2554
2555
2556
2557
2558
2559
2560
2561
2562
2563
2564
2565
2566
2567
2568
2569
2570
2571
2572
2573
2574
2575
2576
2577
2578
2579
2580
2581
2582
2583
2584
2585
2586
2587
2588
2589
2590
2591
2592
2593
2594
2595
2596
2597
2598
2599
2600
2601
2602
2603
2604
2605
2606
2607
2608
2609
2610
2611
2612
2613
2614
2615
2616
2617
2618
2619
2620
2621
2622
2623
2624
2625
2626
2627
2628
2629
2630
2631
2632
2633
2634
2635
2636
2637
2638
2639
2640
2641
2642
2643
2644
2645
2646
2647
2648
2649
2650
2651
2652
2653
2654
2655
2656
2657
2658
2659
2660
2661
2662
2663
2664
2665
2666
2667
2668
2669
2670
2671
2672
2673
2674
2675
2676
2677
2678
2679
2680
2681
2682
2683
2684
2685
2686
2687
2688
2689
2690
2691
2692
2693
2694
2695
2696
2697
2698
2699
2700
2701
2702
2703
2704
2705
2706
2707
2708
2709
2710
2711
2712
2713
2714
2715
2716
2717
2718
2719
2720
2721
2722
2723
2724
2725
2726
2727
2728
2729
2730
2731
2732
2733
2734
2735
2736
2737
2738
2739
2740
2741
2742
2743
2744
2745
2746
2747
2748
2749
2750
2751
2752
2753
2754
2755
2756
2757
2758
2759
2760
2761
2762
2763
2764
2765
2766
2767
2768
2769
2770
2771
2772
2773
2774
2775
2776
2777
2778
2779
2780
2781
2782
2783
2784
2785
2786
2787
2788
2789
2790
2791
2792
2793
2794
2795
2796
2797
2798
2799
2800
2801
2802
2803
2804
2805
2806
2807
2808
2809
2810
2811
2812
2813
2814
2815
2816
2817
2818
2819
2820
2821
2822
2823
2824
2825
2826
2827
2828
2829
2830
2831
2832
2833
2834
2835
2836
2837
2838
2839
2840
2841
2842
2843
2844
2845
2846
2847
2848
2849
2850
2851
2852
2853
2854
2855
2856
2857
2858
2859
2860
2861
2862
2863
2864
2865
2866
2867
2868
2869
2870
2871
2872
2873
2874
2875
2876
2877
2878
2879
2880
2881
2882
2883
2884
2885
2886
2887
2888
2889
2890
2891
2892
2893
2894
2895
2896
2897
2898
2899
2900
2901
2902
2903
2904
2905
2906
2907
2908
2909
2910
2911
2912
2913
2914
2915
2916
2917
2918
2919
2920
2921
2922
2923
2924
2925
2926
2927
2928
2929
2930
2931
2932
2933
2934
2935
2936
2937
2938
2939
2940
2941
2942
2943
2944
2945
2946
2947
2948
2949
2950
2951
2952
2953
2954
2955
2956
2957
2958
2959
2960
2961
2962
2963
2964
2965
2966
2967
2968
2969
2970
2971
2972
2973
2974
2975
2976
2977
2978
2979
2980
2981
2982
2983
2984
2985
2986
2987
2988
2989
2990
2991
2992
2993
2994
2995
2996
2997
2998
2999
3000
3001
3002
3003
3004
3005
3006
3007
3008
3009
3010
3011
3012
3013
3014
3015
3016
3017
3018
3019
3020
3021
3022
3023
3024
3025
3026
3027
3028
3029
3030
3031
3032
3033
3034
3035
3036
3037
3038
3039
3040
3041
3042
3043
3044
3045
3046
3047
3048
3049
3050
3051
3052
3053
3054
3055
3056
3057
3058
3059
3060
3061
3062
3063
3064
3065
3066
3067
3068
3069
3070
3071
3072
3073
3074
3075
3076
3077
3078
3079
3080
3081
3082
3083
3084
3085
3086
3087
3088
3089
3090
3091
3092
3093
3094
3095
3096
3097
3098
3099
3100
3101
3102
3103
3104
3105
3106
3107
3108
3109
3110
3111
3112
3113
3114
3115
3116
3117
3118
3119
3120
3121
3122
3123
3124
3125
3126
3127
3128
3129
3130
3131
3132
3133
3134
3135
3136
3137
3138
3139
3140
3141
3142
3143
3144
3145
3146
3147
3148
3149
3150
3151
3152
3153
3154
3155
3156
3157
3158
3159
3160
3161
3162
3163
3164
3165
3166
3167
3168
3169
3170
3171
3172
3173
3174
3175
3176
3177
3178
3179
3180
3181
3182
3183
3184
3185
3186
3187
3188
3189
3190
3191
3192
3193
3194
3195
3196
3197
3198
3199
3200
3201
3202
3203
3204
3205
3206
3207
3208
3209
3210
3211
3212
3213
3214
3215
3216
3217
3218
3219
3220
3221
3222
3223
3224
3225
3226
3227
3228
3229
3230
3231
3232
3233
3234
3235
3236
3237
3238
3239
3240
3241
3242
3243
3244
3245
3246
3247
3248
3249
3250
3251
3252
3253
3254
3255
3256
3257
3258
3259
3260
3261
3262
3263
3264
3265
3266
3267
3268
3269
3270
3271
3272
3273
3274
3275
3276
3277
3278
3279
3280
3281
3282
3283
3284
3285
3286
3287
3288
3289
3290
3291
3292
3293
3294
3295
3296
3297
3298
3299
3300
3301
3302
3303
3304
3305
3306
3307
3308
3309
3310
3311
3312
3313
3314
3315
3316
3317
3318
3319
3320
3321
3322
3323
3324
3325
3326
3327
3328
3329
3330
3331
3332
3333
3334
3335
3336
3337
3338
3339
3340
3341
3342
3343
3344
3345
3346
3347
3348
3349
3350
3351
3352
3353
3354
3355
3356
3357
3358
3359
3360
3361
3362
3363
3364
3365
3366
3367
3368
3369
3370
3371
3372
3373
3374
3375
3376
3377
3378
3379
3380
3381
3382
3383
3384
3385
3386
3387
3388
3389
3390
3391
3392
3393
3394
3395
3396
3397
3398
3399
3400
3401
3402
3403
3404
3405
3406
3407
3408
3409
3410
3411
3412
3413
3414
3415
3416
3417
3418
3419
3420
3421
3422
3423
3424
3425
3426
3427
3428
3429
3430
3431
3432
3433
3434
3435
3436
3437
3438
3439
3440
3441
3442
3443
3444
3445
3446
3447
3448
3449
3450
3451
3452
3453
3454
3455
3456
3457
3458
3459
3460
3461
3462
3463
3464
3465
3466
3467
3468
3469
3470
3471
3472
3473
3474
3475
3476
3477
3478
3479
3480
3481
3482
3483
3484
3485
3486
3487
3488
3489
3490
3491
3492
3493
3494
3495
3496
3497
3498
3499
3500
3501
3502
3503
3504
3505
3506
3507
3508
3509
3510
3511
3512
3513
3514
3515
3516
3517
3518
3519
3520
3521
3522
3523
3524
3525
3526
3527
3528
3529
3530
3531
3532
3533
3534
3535
3536
3537
3538
3539
3540
3541
3542
3543
3544
3545
3546
3547
3548
3549
3550
3551
3552
3553
3554
3555
3556
3557
3558
3559
3560
3561
3562
3563
3564
3565
3566
3567
3568
3569
3570
3571
3572
3573
3574
3575
3576
3577
3578
3579
3580
3581
3582
3583
3584
3585
3586
3587
3588
3589
3590
3591
3592
3593
3594
3595
3596
3597
3598
3599
3600
3601
3602
3603
3604
3605
3606
3607
3608
3609
3610
3611
3612
3613
3614
3615
3616
3617
3618
3619
3620
3621
3622
3623
3624
3625
3626
3627
3628
3629
3630
3631
3632
3633
3634
3635
3636
3637
3638
3639
3640
3641
3642
3643
3644
3645
3646
3647
3648
3649
3650
3651
3652
3653
3654
3655
3656
3657
3658
3659
3660
3661
3662
3663
3664
3665
3666
3667
3668
3669
3670
3671
3672
3673
3674
3675
3676
3677
3678
3679
3680
3681
3682
3683
3684
3685
3686
3687
3688
3689
3690
3691
3692
3693
3694
3695
3696
3697
3698
3699
3700
3701
3702
3703
3704
3705
3706
3707
3708
3709
3710
3711
3712
3713
3714
3715
3716
3717
3718
3719
3720
3721
3722
3723
3724
3725
3726
3727
3728
3729
3730
3731
3732
3733
3734
3735
3736
3737
3738
3739
3740
3741
3742
3743
3744
3745
3746
3747
3748
3749
3750
3751
3752
3753
3754
3755
3756
3757
3758
3759
3760
3761
3762
3763
3764
3765
3766
3767
3768
3769
3770
3771
3772
3773
3774
3775
3776
3777
3778
3779
3780
3781
3782
3783
3784
3785
3786
3787
3788
3789
3790
3791
3792
3793
3794
3795
3796
3797
3798
3799
3800
3801
3802
3803
3804
3805
3806
3807
3808
3809
3810
3811
3812
3813
3814
3815
3816
3817
3818
3819
3820
3821
3822
3823
3824
3825
3826
3827
3828
3829
3830
3831
3832
3833
3834
3835
3836
3837
3838
3839
3840
3841
3842
3843
3844
3845
3846
3847
3848
3849
3850
3851
3852
3853
3854
3855
3856
3857
3858
3859
3860
3861
3862
3863
3864
3865
3866
3867
3868
3869
3870
3871
3872
3873
3874
3875
3876
3877
3878
3879
3880
3881
3882
3883
3884
3885
3886
3887
3888
3889
3890
3891
3892
3893
3894
3895
3896
3897
3898
3899
3900
3901
3902
3903
3904
3905
3906
3907
3908
3909
3910
3911
3912
3913
3914
3915
3916
3917
3918
3919
3920
3921
3922
3923
3924
3925
3926
3927
3928
3929
3930
3931
3932
3933
3934
3935
3936
3937
3938
3939
3940
3941
3942
3943
3944
3945
3946
3947
3948
3949
3950
3951
3952
3953
3954
3955
3956
3957
3958
3959
3960
3961
3962
3963
3964
3965
3966
3967
3968
3969
3970
3971
3972
3973
3974
3975
3976
3977
3978
3979
3980
3981
3982
3983
3984
3985
3986
3987
3988
3989
3990
3991
3992
3993
3994
3995
3996
3997
3998
3999
4000
4001
4002
4003
4004
4005
4006
4007
4008
4009
4010
4011
4012
4013
4014
4015
4016
4017
4018
4019
4020
4021
4022
4023
4024
4025
4026
4027
4028
4029
4030
4031
4032
4033
4034
4035
4036
4037
4038
4039
4040
4041
4042
4043
4044
4045
4046
4047
4048
4049
4050
4051
4052
4053
4054
4055
4056
4057
4058
4059
4060
4061
4062
4063
4064
4065
4066
4067
4068
4069
4070
4071
4072
4073
4074
4075
4076
4077
4078
4079
4080
4081
4082
4083
4084
4085
4086
4087
4088
4089
4090
4091
4092
4093
4094
4095
4096
4097
4098
4099
4100
4101
4102
4103
4104
4105
4106
4107
4108
4109
4110
4111
4112
4113
4114
4115
4116
4117
4118
4119
4120
4121
4122
4123
4124
4125
4126
4127
4128
4129
4130
4131
4132
4133
4134
4135
4136
4137
4138
4139
4140
4141
4142
4143
4144
4145
4146
4147
4148
4149
4150
4151
4152
4153
4154
4155
4156
4157
4158
4159
4160
4161
4162
4163
4164
4165
4166
4167
4168
4169
4170
4171
4172
4173
4174
4175
4176
4177
4178
4179
4180
4181
4182
4183
4184
4185
4186
4187
4188
4189
4190
4191
4192
4193
4194
4195
4196
4197
4198
4199
4200
4201
4202
4203
4204
4205
4206
4207
4208
4209
4210
4211
4212
4213
4214
4215
4216
4217
4218
4219
4220
4221
4222
4223
4224
4225
4226
4227
4228
4229
4230
4231
4232
4233
4234
4235
4236
4237
4238
4239
4240
4241
4242
4243
4244
4245
4246
4247
4248
4249
4250
4251
4252
4253
4254
4255
4256
4257
4258
4259
4260
4261
4262
4263
4264
4265
4266
4267
4268
4269
4270
4271
4272
4273
4274
4275
4276
4277
4278
4279
4280
4281
4282
4283
4284
4285
4286
4287
4288
4289
4290
4291
4292
4293
4294
4295
4296
4297
4298
4299
4300
4301
4302
4303
4304
4305
4306
4307
4308
4309
4310
4311
4312
4313
4314
4315
4316
4317
4318
4319
4320
4321
4322
4323
4324
4325
4326
4327
4328
4329
4330
4331
4332
4333
4334
4335
4336
4337
4338
4339
4340
4341
4342
4343
4344
4345
4346
4347
4348
4349
4350
4351
4352
4353
4354
4355
4356
4357
4358
4359
4360
4361
4362
4363
4364
4365
4366
4367
4368
4369
4370
4371
4372
4373
4374
4375
4376
4377
4378
4379
4380
4381
4382
4383
4384
4385
4386
4387
4388
4389
4390
4391
4392
4393
4394
4395
4396
4397
4398
4399
4400
4401
4402
4403
4404
4405
4406
4407
4408
4409
4410
4411
4412
4413
4414
4415
4416
4417
4418
4419
4420
4421
4422
4423
4424
4425
4426
4427
4428
4429
4430
4431
4432
4433
4434
4435
4436
4437
4438
4439
4440
4441
4442
4443
4444
4445
4446
4447
4448
4449
4450
4451
4452
4453
4454
4455
4456
4457
4458
4459
4460
4461
4462
4463
4464
4465
4466
4467
4468
4469
4470
4471
4472
4473
4474
4475
4476
4477
4478
4479
4480
4481
4482
4483
4484
4485
4486
4487
4488
4489
4490
4491
4492
4493
4494
4495
4496
4497
4498
4499
4500
4501
4502
4503
4504
4505
4506
4507
4508
4509
4510
4511
4512
4513
4514
4515
4516
4517
4518
4519
4520
4521
4522
4523
4524
4525
4526
4527
4528
4529
4530
4531
4532
4533
4534
4535
4536
4537
4538
4539
4540
4541
4542
4543
4544
4545
4546
4547
4548
4549
4550
4551
4552
4553
4554
4555
4556
4557
4558
4559
4560
4561
4562
4563
4564
4565
4566
4567
4568
4569
4570
4571
4572
4573
4574
4575
4576
4577
4578
4579
4580
4581
4582
4583
4584
4585
4586
4587
4588
4589
4590
4591
4592
4593
4594
4595
4596
4597
4598
4599
4600
4601
4602
4603
4604
4605
4606
4607
4608
4609
4610
4611
4612
4613
4614
4615
4616
4617
4618
4619
4620
4621
4622
4623
4624
4625
4626
4627
4628
4629
4630
4631
4632
4633
4634
4635
4636
4637
4638
4639
4640
4641
4642
4643
4644
4645
4646
4647
4648
4649
4650
4651
4652
4653
4654
4655
4656
4657
4658
4659
4660
4661
4662
4663
4664
4665
4666
4667
4668
4669
4670
4671
4672
4673
4674
4675
4676
4677
4678
4679
4680
4681
4682
4683
4684
4685
4686
4687
4688
4689
4690
4691
4692
4693
4694
4695
4696
4697
4698
4699
4700
4701
4702
4703
4704
4705
4706
4707
4708
4709
4710
4711
4712
4713
4714
4715
4716
4717
4718
4719
4720
4721
4722
4723
4724
4725
4726
4727
4728
4729
4730
4731
4732
4733
4734
4735
4736
4737
4738
4739
4740
4741
4742
4743
4744
4745
4746
4747
4748
4749
4750
4751
4752
4753
4754
4755
4756
4757
4758
4759
4760
4761
4762
4763
4764
4765
4766
4767
4768
4769
4770
4771
4772
4773
4774
4775
4776
4777
4778
4779
4780
4781
4782
4783
4784
4785
4786
4787
4788
4789
4790
4791
4792
4793
4794
4795
4796
4797
4798
4799
4800
4801
4802
4803
4804
4805
4806
4807
4808
4809
4810
4811
4812
4813
4814
4815
4816
4817
4818
4819
4820
4821
4822
4823
4824
4825
4826
4827
4828
4829
4830
4831
4832
4833
4834
4835
4836
4837
4838
4839
4840
4841
4842
4843
4844
4845
4846
4847
4848
4849
4850
4851
4852
4853
4854
4855
4856
4857
4858
4859
4860
4861
4862
4863
4864
4865
4866
4867
4868
4869
4870
4871
4872
4873
4874
4875
4876
4877
4878
4879
4880
4881
4882
4883
4884
4885
4886
4887
4888
4889
4890
4891
4892
4893
4894
4895
4896
4897
4898
4899
4900
4901
4902
4903
4904
4905
4906
4907
4908
4909
4910
4911
4912
4913
4914
4915
4916
4917
4918
4919
4920
4921
4922
4923
4924
4925
4926
4927
4928
4929
4930
4931
4932
4933
4934
4935
4936
4937
4938
4939
4940
4941
4942
4943
4944
4945
4946
4947
4948
4949
4950
4951
4952
4953
4954
4955
4956
4957
4958
4959
4960
4961
4962
4963
4964
4965
4966
4967
4968
4969
4970
4971
4972
4973
4974
4975
4976
4977
4978
4979
4980
4981
4982
4983
4984
4985
4986
4987
4988
4989
4990
4991
4992
4993
4994
4995
4996
4997
4998
4999
5000
5001
5002
5003
5004
5005
5006
5007
5008
5009
5010
5011
5012
5013
5014
5015
5016
5017
5018
5019
5020
5021
5022
5023
5024
5025
5026
5027
5028
5029
5030
5031
5032
5033
5034
5035
5036
5037
5038
5039
5040
5041
5042
5043
5044
5045
5046
5047
5048
5049
5050
5051
5052
5053
5054
5055
5056
5057
5058
5059
5060
5061
5062
5063
5064
5065
5066
5067
5068
5069
5070
5071
5072
5073
5074
5075
5076
5077
5078
5079
5080
5081
5082
5083
5084
5085
5086
5087
5088
5089
5090
5091
5092
5093
5094
5095
5096
5097
5098
5099
5100
5101
5102
5103
5104
5105
5106
5107
5108
5109
5110
5111
5112
5113
5114
5115
5116
5117
5118
5119
5120
5121
5122
5123
5124
5125
5126
5127
5128
5129
5130
5131
5132
5133
5134
5135
5136
5137
5138
5139
5140
5141
5142
5143
5144
5145
5146
5147
5148
5149
5150
5151
5152
5153
5154
5155
5156
5157
5158
5159
5160
5161
5162
5163
5164
5165
5166
5167
5168
5169
5170
5171
5172
5173
5174
5175
5176
5177
5178
5179
5180
5181
5182
5183
5184
5185
5186
5187
5188
5189
5190
5191
5192
5193
5194
5195
5196
5197
5198
5199
5200
5201
5202
5203
5204
5205
5206
5207
5208
5209
5210
5211
5212
5213
5214
5215
5216
5217
5218
5219
5220
5221
5222
5223
5224
5225
5226
5227
5228
5229
5230
5231
5232
5233
5234
5235
5236
5237
5238
5239
5240
5241
5242
5243
5244
5245
5246
5247
5248
5249
5250
5251
5252
5253
5254
5255
5256
5257
5258
5259
5260
5261
5262
5263
5264
5265
5266
5267
5268
5269
5270
5271
5272
5273
5274
5275
5276
5277
5278
5279
5280
5281
5282
5283
5284
5285
5286
5287
5288
5289
5290
5291
5292
5293
5294
5295
5296
5297
5298
5299
5300
5301
5302
5303
5304
5305
5306
5307
5308
5309
5310
5311
5312
5313
5314
5315
5316
5317
5318
5319
5320
5321
5322
5323
5324
5325
5326
5327
5328
5329
5330
5331
5332
5333
5334
5335
5336
5337
5338
5339
5340
5341
5342
5343
5344
5345
5346
5347
5348
5349
5350
5351
5352
5353
5354
5355
5356
5357
5358
5359
5360
5361
5362
5363
5364
5365
5366
5367
5368
5369
5370
5371
5372
5373
5374
5375
5376
5377
5378
5379
5380
5381
5382
5383
5384
5385
5386
5387
5388
5389
5390
5391
5392
5393
5394
5395
5396
5397
5398
5399
5400
5401
5402
5403
5404
5405
5406
5407
5408
5409
5410
5411
5412
5413
5414
5415
5416
5417
5418
5419
5420
5421
5422
5423
5424
5425
5426
5427
5428
5429
5430
5431
5432
5433
5434
5435
5436
5437
5438
5439
5440
5441
5442
5443
5444
5445
5446
5447
5448
5449
5450
5451
5452
5453
5454
5455
5456
5457
5458
5459
5460
5461
5462
5463
5464
5465
5466
5467
5468
5469
5470
5471
5472
5473
5474
5475
5476
5477
5478
5479
5480
5481
5482
5483
5484
5485
5486
5487
5488
5489
5490
5491
5492
5493
5494
5495
5496
5497
5498
5499
5500
5501
5502
5503
5504
5505
5506
5507
5508
5509
5510
5511
5512
5513
5514
5515
5516
5517
5518
5519
5520
5521
5522
5523
5524
5525
5526
5527
5528
5529
5530
5531
5532
5533
5534
5535
5536
5537
5538
5539
5540
5541
5542
5543
5544
5545
5546
5547
5548
5549
5550
5551
5552
5553
5554
5555
5556
5557
5558
5559
5560
5561
5562
5563
5564
5565
5566
5567
5568
5569
5570
5571
5572
5573
5574
5575
5576
5577
5578
5579
5580
5581
5582
5583
5584
5585
5586
5587
5588
5589
5590
5591
5592
5593
5594
5595
5596
5597
5598
5599
5600
5601
5602
5603
5604
5605
5606
5607
5608
5609
5610
5611
5612
5613
5614
5615
5616
5617
5618
5619
5620
5621
5622
5623
5624
5625
5626
5627
5628
5629
5630
5631
5632
5633
5634
5635
5636
5637
5638
5639
5640
5641
5642
5643
5644
5645
5646
5647
5648
5649
5650
5651
5652
5653
5654
5655
5656
5657
5658
5659
5660
5661
5662
5663
5664
5665
5666
5667
5668
5669
5670
5671
5672
5673
5674
5675
5676
5677
5678
5679
5680
5681
5682
5683
5684
5685
5686
5687
5688
5689
5690
5691
5692
5693
5694
5695
5696
5697
5698
5699
5700
5701
5702
5703
5704
5705
5706
5707
5708
5709
5710
5711
5712
5713
5714
5715
5716
5717
5718
5719
5720
5721
5722
5723
5724
5725
5726
5727
5728
5729
5730
5731
5732
5733
5734
5735
5736
5737
5738
5739
5740
5741
5742
5743
5744
5745
5746
5747
5748
5749
5750
5751
5752
5753
5754
5755
5756
5757
5758
5759
5760
5761
5762
5763
5764
5765
5766
5767
5768
5769
5770
5771
5772
5773
5774
5775
5776
5777
5778
5779
5780
5781
5782
5783
5784
5785
5786
5787
5788
5789
5790
5791
5792
5793
5794
5795
5796
5797
5798
5799
5800
5801
5802
5803
5804
5805
5806
5807
5808
5809
5810
5811
5812
5813
5814
5815
5816
5817
5818
5819
5820
5821
5822
5823
5824
5825
5826
5827
5828
5829
5830
5831
5832
5833
5834
5835
5836
5837
5838
5839
5840
5841
5842
5843
5844
5845
5846
5847
5848
5849
5850
5851
5852
5853
5854
5855
5856
5857
5858
5859
5860
5861
5862
5863
5864
5865
5866
5867
5868
5869
5870
5871
5872
5873
5874
5875
5876
5877
5878
5879
5880
5881
5882
5883
5884
5885
5886
5887
5888
5889
5890
5891
5892
5893
5894
5895
5896
5897
5898
5899
5900
5901
5902
5903
5904
5905
5906
5907
5908
5909
5910
5911
5912
5913
5914
5915
5916
5917
5918
5919
5920
5921
5922
5923
5924
5925
5926
5927
5928
5929
5930
5931
5932
5933
5934
5935
5936
5937
5938
5939
5940
5941
5942
5943
5944
5945
5946
5947
5948
5949
5950
5951
5952
5953
5954
5955
5956
5957
5958
5959
5960
5961
5962
5963
5964
5965
5966
5967
5968
5969
5970
5971
5972
5973
5974
5975
5976
5977
5978
5979
5980
5981
5982
5983
5984
5985
5986
5987
5988
5989
5990
5991
5992
5993
5994
5995
5996
5997
5998
5999
6000
6001
6002
6003
6004
6005
6006
6007
6008
6009
6010
6011
6012
6013
6014
6015
6016
6017
6018
6019
6020
6021
6022
6023
6024
6025
6026
6027
6028
6029
6030
6031
6032
6033
6034
6035
6036
6037
6038
6039
6040
6041
6042
6043
6044
6045
6046
6047
6048
6049
6050
6051
6052
6053
6054
6055
6056
6057
6058
6059
6060
6061
6062
6063
6064
6065
6066
6067
6068
6069
6070
6071
6072
6073
6074
6075
6076
6077
6078
6079
6080
6081
6082
6083
6084
6085
6086
6087
6088
6089
6090
6091
6092
6093
6094
6095
6096
6097
6098
6099
6100
6101
6102
6103
6104
6105
6106
6107
6108
6109
6110
6111
6112
6113
6114
6115
6116
6117
6118
6119
6120
6121
6122
6123
6124
6125
6126
6127
6128
6129
6130
6131
6132
6133
6134
6135
6136
6137
6138
6139
6140
6141
6142
6143
6144
6145
6146
6147
6148
6149
6150
6151
6152
6153
6154
6155
6156
6157
6158
6159
6160
6161
6162
6163
6164
6165
6166
6167
6168
6169
6170
6171
6172
6173
6174
6175
6176
6177
6178
6179
6180
6181
6182
6183
6184
6185
6186
6187
6188
6189
6190
6191
6192
6193
6194
6195
6196
6197
6198
6199
6200
6201
6202
6203
6204
6205
6206
6207
6208
6209
6210
6211
6212
6213
6214
6215
6216
6217
6218
6219
6220
6221
6222
6223
6224
6225
6226
6227
6228
6229
6230
6231
6232
6233
6234
6235
6236
6237
6238
6239
6240
6241
6242
6243
6244
6245
6246
6247
6248
6249
6250
6251
6252
6253
6254
6255
6256
6257
6258
6259
6260
6261
6262
6263
6264
6265
6266
6267
6268
6269
6270
6271
6272
6273
6274
6275
6276
6277
6278
6279
6280
6281
6282
6283
6284
6285
6286
6287
6288
6289
6290
6291
6292
6293
6294
6295
6296
6297
6298
6299
6300
6301
6302
6303
6304
6305
6306
6307
6308
6309
6310
6311
6312
6313
6314
6315
6316
6317
6318
6319
6320
6321
6322
6323
6324
6325
6326
6327
6328
6329
6330
6331
6332
6333
6334
6335
6336
6337
6338
6339
6340
6341
6342
6343
6344
6345
6346
6347
6348
6349
6350
6351
6352
6353
6354
6355
6356
6357
6358
6359
6360
6361
6362
6363
6364
6365
6366
6367
6368
6369
6370
6371
6372
6373
6374
6375
6376
6377
6378
6379
6380
6381
6382
6383
6384
6385
6386
6387
6388
6389
6390
6391
6392
6393
6394
6395
6396
6397
6398
6399
6400
6401
6402
6403
6404
6405
6406
6407
6408
6409
6410
6411
6412
6413
6414
6415
6416
6417
6418
6419
6420
6421
6422
6423
6424
6425
6426
6427
6428
6429
6430
6431
6432
6433
6434
6435
6436
6437
6438
6439
6440
6441
6442
6443
6444
6445
6446
6447
6448
6449
6450
6451
6452
6453
6454
6455
6456
6457
6458
6459
6460
6461
6462
6463
6464
6465
6466
6467
6468
6469
6470
6471
6472
6473
6474
6475
6476
6477
6478
6479
6480
6481
6482
6483
6484
6485
6486
6487
6488
6489
6490
6491
6492
6493
6494
6495
6496
6497
6498
6499
6500
6501
6502
6503
6504
6505
6506
6507
6508
6509
6510
6511
6512
6513
6514
6515
6516
6517
6518
6519
6520
6521
6522
6523
6524
6525
6526
6527
6528
6529
6530
6531
6532
6533
6534
6535
6536
6537
6538
6539
6540
6541
6542
6543
6544
6545
6546
6547
6548
6549
6550
6551
6552
6553
6554
6555
6556
6557
6558
6559
6560
6561
6562
6563
6564
6565
6566
6567
6568
6569
6570
6571
6572
6573
6574
6575
6576
6577
6578
6579
6580
6581
6582
6583
6584
6585
6586
6587
6588
6589
6590
6591
6592
6593
6594
6595
6596
6597
6598
6599
6600
6601
6602
6603
6604
6605
6606
6607
6608
6609
6610
6611
6612
6613
6614
6615
6616
6617
6618
6619
6620
6621
6622
6623
6624
6625
6626
6627
6628
6629
6630
6631
6632
6633
6634
6635
6636
6637
6638
6639
6640
6641
6642
6643
6644
6645
6646
6647
6648
6649
6650
6651
6652
6653
6654
6655
6656
6657
6658
6659
6660
6661
6662
6663
6664
6665
6666
6667
6668
6669
6670
6671
6672
6673
6674
6675
6676
6677
6678
6679
6680
6681
6682
6683
6684
6685
6686
6687
6688
6689
6690
6691
6692
6693
6694
6695
6696
6697
6698
6699
6700
6701
6702
6703
6704
6705
6706
6707
6708
6709
6710
6711
6712
6713
6714
6715
6716
6717
6718
6719
6720
6721
6722
6723
6724
6725
6726
6727
6728
6729
6730
6731
6732
6733
6734
6735
6736
6737
6738
6739
6740
6741
6742
6743
6744
6745
6746
6747
6748
6749
6750
6751
6752
6753
6754
6755
6756
6757
6758
6759
6760
6761
6762
6763
6764
6765
6766
6767
6768
6769
6770
6771
6772
6773
6774
6775
6776
6777
6778
6779
6780
6781
6782
6783
6784
6785
6786
6787
6788
6789
6790
6791
6792
6793
6794
6795
6796
6797
6798
6799
6800
6801
6802
6803
6804
6805
6806
6807
6808
6809
6810
6811
6812
6813
6814
6815
6816
6817
6818
6819
6820
6821
6822
6823
6824
6825
6826
6827
6828
6829
6830
6831
6832
6833
6834
6835
6836
6837
6838
6839
6840
6841
6842
6843
6844
6845
6846
6847
6848
6849
6850
6851
6852
6853
6854
6855
6856
6857
6858
6859
6860
6861
6862
6863
6864
6865
6866
6867
6868
6869
6870
6871
6872
6873
6874
6875
6876
6877
6878
6879
6880
6881
6882
6883
6884
6885
6886
6887
6888
6889
6890
6891
6892
6893
6894
6895
6896
6897
6898
6899
6900
6901
6902
6903
6904
6905
6906
6907
6908
6909
6910
6911
6912
6913
6914
6915
6916
6917
6918
6919
6920
6921
6922
6923
6924
6925
6926
6927
6928
6929
6930
6931
6932
6933
6934
6935
6936
6937
6938
6939
6940
6941
6942
6943
6944
6945
6946
6947
6948
6949
6950
6951
6952
6953
6954
6955
6956
6957
6958
6959
6960
6961
6962
6963
6964
6965
6966
6967
6968
6969
6970
6971
6972
6973
6974
6975
6976
6977
6978
6979
6980
6981
6982
6983
6984
6985
6986
6987
6988
6989
6990
6991
6992
6993
6994
6995
6996
6997
6998
6999
7000
7001
7002
7003
7004
7005
7006
7007
7008
7009
7010
7011
7012
7013
7014
7015
7016
7017
7018
7019
7020
7021
7022
7023
7024
7025
7026
7027
7028
7029
7030
7031
7032
7033
7034
7035
7036
7037
7038
7039
7040
7041
7042
7043
7044
7045
7046
7047
7048
7049
7050
7051
7052
7053
7054
7055
7056
7057
7058
7059
7060
7061
7062
7063
7064
7065
7066
7067
7068
7069
7070
7071
7072
7073
7074
7075
7076
7077
7078
7079
7080
7081
7082
7083
7084
7085
7086
7087
7088
7089
7090
7091
7092
7093
7094
7095
7096
7097
7098
7099
7100
7101
7102
7103
7104
7105
7106
7107
7108
7109
7110
7111
7112
7113
7114
7115
7116
7117
7118
7119
7120
7121
7122
7123
7124
7125
7126
7127
7128
7129
7130
7131
7132
7133
7134
7135
7136
7137
7138
7139
7140
7141
7142
7143
7144
7145
7146
7147
7148
7149
7150
7151
7152
7153
7154
7155
7156
7157
7158
7159
7160
7161
7162
7163
7164
7165
7166
7167
7168
7169
7170
7171
7172
7173
7174
7175
7176
7177
7178
7179
7180
7181
7182
7183
7184
7185
7186
7187
7188
7189
7190
7191
7192
7193
7194
7195
7196
7197
7198
7199
7200
7201
7202
7203
7204
7205
7206
7207
7208
7209
7210
7211
7212
7213
7214
7215
7216
7217
7218
7219
7220
7221
7222
7223
7224
7225
7226
7227
7228
7229
7230
7231
7232
7233
7234
7235
7236
7237
7238
7239
7240
7241
7242
7243
7244
7245
7246
7247
7248
7249
7250
7251
7252
7253
7254
7255
7256
7257
7258
7259
7260
7261
7262
7263
7264
7265
7266
7267
7268
7269
7270
7271
7272
7273
7274
7275
7276
7277
7278
7279
7280
7281
7282
7283
7284
7285
7286
7287
7288
7289
7290
7291
7292
7293
7294
7295
7296
7297
7298
7299
7300
7301
7302
7303
7304
7305
7306
7307
7308
7309
7310
7311
7312
7313
7314
7315
7316
7317
7318
7319
7320
7321
7322
7323
7324
7325
7326
7327
7328
7329
7330
7331
7332
7333
7334
7335
7336
7337
7338
7339
7340
7341
7342
7343
7344
7345
7346
7347
7348
7349
7350
7351
7352
7353
7354
7355
7356
7357
7358
7359
7360
7361
7362
7363
7364
7365
7366
7367
7368
7369
7370
7371
7372
7373
7374
7375
7376
7377
7378
7379
7380
7381
7382
7383
7384
7385
7386
7387
7388
7389
7390
7391
7392
7393
7394
7395
7396
7397
7398
7399
7400
7401
7402
7403
7404
7405
7406
7407
7408
7409
7410
7411
7412
7413
7414
7415
7416
7417
7418
7419
7420
7421
7422
7423
7424
7425
7426
7427
7428
7429
7430
7431
7432
7433
7434
7435
7436
7437
7438
7439
7440
7441
7442
7443
7444
7445
7446
7447
7448
7449
7450
7451
7452
7453
7454
7455
7456
7457
7458
7459
7460
7461
7462
7463
7464
7465
7466
7467
7468
7469
7470
7471
7472
7473
7474
7475
7476
7477
7478
7479
7480
7481
7482
7483
7484
7485
7486
7487
7488
7489
7490
7491
7492
7493
7494
7495
7496
7497
7498
7499
7500
7501
7502
7503
7504
7505
7506
7507
7508
7509
7510
7511
7512
7513
7514
7515
7516
7517
7518
7519
7520
7521
7522
7523
7524
7525
7526
7527
7528
7529
7530
7531
7532
7533
7534
7535
7536
7537
7538
7539
7540
7541
7542
7543
7544
7545
7546
7547
7548
7549
7550
7551
7552
7553
7554
7555
7556
7557
7558
7559
7560
7561
7562
7563
7564
7565
7566
7567
7568
7569
7570
7571
7572
7573
7574
7575
7576
7577
7578
7579
7580
7581
7582
7583
7584
7585
7586
7587
7588
7589
7590
7591
7592
7593
7594
7595
7596
7597
7598
7599
7600
7601
7602
7603
7604
7605
7606
7607
7608
7609
7610
7611
7612
7613
7614
7615
7616
7617
7618
7619
7620
7621
7622
7623
7624
7625
7626
7627
7628
7629
7630
7631
7632
7633
7634
7635
7636
7637
7638
7639
7640
7641
7642
7643
7644
7645
7646
7647
7648
7649
7650
7651
7652
7653
7654
7655
7656
7657
7658
7659
7660
7661
7662
7663
7664
7665
7666
7667
7668
7669
7670
7671
7672
7673
7674
7675
7676
7677
7678
7679
7680
7681
7682
7683
7684
7685
7686
7687
7688
7689
7690
7691
7692
7693
7694
7695
7696
7697
7698
7699
7700
7701
7702
7703
7704
7705
7706
7707
7708
7709
7710
7711
7712
7713
7714
7715
7716
7717
7718
7719
7720
7721
7722
7723
7724
7725
7726
7727
7728
7729
7730
7731
7732
7733
7734
7735
7736
7737
7738
7739
7740
7741
7742
7743
7744
7745
7746
7747
7748
7749
7750
7751
7752
7753
7754
7755
7756
7757
7758
7759
7760
7761
7762
7763
7764
7765
7766
7767
7768
7769
7770
7771
7772
7773
7774
7775
7776
7777
7778
7779
7780
7781
7782
7783
7784
7785
7786
7787
7788
7789
7790
7791
7792
7793
7794
7795
7796
7797
7798
7799
7800
7801
7802
7803
7804
7805
7806
7807
7808
7809
7810
7811
7812
7813
7814
7815
7816
7817
7818
7819
7820
7821
7822
7823
7824
7825
7826
7827
7828
7829
7830
7831
7832
7833
7834
7835
7836
7837
7838
7839
7840
7841
7842
7843
7844
7845
7846
7847
7848
7849
7850
7851
7852
7853
7854
7855
7856
7857
7858
7859
7860
7861
7862
7863
7864
7865
7866
7867
7868
7869
7870
7871
7872
7873
7874
7875
7876
7877
7878
7879
7880
7881
7882
7883
7884
7885
7886
7887
7888
7889
7890
7891
7892
7893
7894
7895
7896
7897
7898
7899
7900
7901
7902
7903
7904
7905
7906
7907
7908
7909
7910
7911
7912
7913
7914
7915
7916
7917
7918
7919
7920
7921
7922
7923
7924
7925
7926
7927
7928
7929
7930
7931
7932
7933
7934
7935
7936
7937
7938
7939
7940
7941
7942
7943
7944
7945
7946
7947
7948
7949
7950
7951
7952
7953
7954
7955
7956
7957
7958
7959
7960
7961
7962
7963
7964
7965
7966
7967
7968
7969
7970
7971
7972
7973
7974
7975
7976
7977
7978
7979
7980
7981
7982
7983
7984
7985
7986
7987
7988
7989
7990
7991
7992
7993
7994
7995
7996
7997
7998
7999
8000
8001
8002
8003
8004
8005
8006
8007
8008
8009
8010
8011
8012
8013
8014
8015
8016
8017
8018
8019
8020
8021
8022
8023
8024
8025
8026
8027
8028
8029
8030
8031
8032
8033
8034
8035
8036
8037
8038
8039
8040
8041
8042
8043
8044
8045
8046
8047
8048
8049
8050
8051
8052
8053
8054
8055
8056
8057
8058
8059
8060
8061
8062
8063
8064
8065
8066
8067
8068
8069
8070
8071
8072
8073
8074
8075
8076
8077
8078
8079
8080
8081
8082
8083
8084
8085
8086
8087
8088
8089
8090
8091
8092
8093
8094
8095
8096
8097
8098
8099
8100
8101
8102
8103
8104
8105
8106
8107
8108
8109
8110
8111
8112
8113
8114
8115
8116
8117
8118
8119
8120
8121
8122
8123
8124
8125
8126
8127
8128
8129
8130
8131
8132
8133
8134
8135
8136
8137
8138
8139
8140
8141
8142
8143
8144
8145
8146
8147
8148
8149
8150
8151
8152
8153
8154
8155
8156
8157
8158
8159
8160
8161
8162
8163
8164
8165
8166
8167
8168
8169
8170
8171
8172
8173
8174
8175
8176
8177
8178
8179
8180
8181
8182
8183
8184
8185
8186
8187
8188
8189
8190
8191
8192
8193
8194
8195
8196
8197
8198
8199
8200
8201
8202
8203
8204
8205
8206
8207
8208
8209
8210
8211
8212
8213
8214
8215
8216
8217
8218
8219
8220
8221
8222
8223
8224
8225
8226
8227
8228
8229
8230
8231
8232
8233
8234
8235
8236
8237
8238
8239
8240
8241
8242
8243
8244
8245
8246
8247
8248
8249
8250
8251
8252
8253
8254
8255
8256
8257
8258
8259
8260
8261
8262
8263
8264
8265
8266
8267
8268
8269
8270
8271
8272
8273
8274
8275
8276
8277
8278
8279
8280
8281
8282
8283
8284
8285
8286
8287
8288
8289
8290
8291
8292
8293
8294
8295
8296
8297
8298
8299
8300
8301
8302
8303
8304
8305
8306
8307
8308
8309
8310
8311
8312
8313
8314
8315
8316
8317
8318
8319
8320
8321
8322
8323
8324
8325
8326
8327
8328
8329
8330
8331
8332
8333
8334
8335
8336
8337
8338
8339
8340
8341
8342
8343
8344
8345
8346
8347
8348
8349
8350
8351
8352
8353
8354
8355
8356
8357
8358
8359
8360
8361
8362
8363
8364
8365
8366
8367
8368
8369
8370
8371
8372
8373
8374
8375
8376
8377
8378
8379
8380
8381
8382
8383
8384
8385
8386
8387
8388
8389
8390
8391
8392
8393
8394
8395
8396
8397
8398
8399
8400
8401
8402
8403
8404
8405
8406
8407
8408
8409
8410
8411
8412
8413
8414
8415
8416
8417
8418
8419
8420
8421
8422
8423
8424
8425
8426
8427
8428
8429
8430
8431
8432
8433
8434
8435
8436
8437
8438
8439
8440
8441
8442
8443
8444
8445
8446
8447
8448
8449
8450
8451
8452
8453
8454
8455
8456
8457
8458
8459
8460
8461
8462
8463
8464
8465
8466
8467
8468
8469
8470
8471
8472
8473
8474
8475
8476
8477
8478
8479
8480
8481
8482
8483
8484
8485
8486
8487
8488
8489
8490
8491
8492
8493
8494
8495
8496
8497
8498
8499
8500
8501
8502
8503
8504
8505
8506
8507
8508
8509
8510
8511
8512
8513
8514
8515
8516
8517
8518
8519
8520
8521
8522
8523
8524
8525
8526
8527
8528
8529
8530
8531
8532
8533
8534
8535
8536
8537
8538
8539
8540
8541
8542
8543
8544
8545
8546
8547
8548
8549
8550
8551
8552
8553
8554
8555
8556
8557
8558
8559
8560
8561
8562
8563
8564
8565
8566
8567
8568
8569
8570
8571
8572
8573
8574
8575
8576
8577
8578
8579
8580
8581
8582
8583
8584
8585
8586
8587
8588
8589
8590
8591
8592
8593
8594
8595
8596
8597
8598
8599
8600
8601
8602
8603
8604
8605
8606
8607
8608
8609
8610
8611
8612
8613
8614
8615
8616
8617
8618
8619
8620
8621
8622
8623
8624
8625
8626
8627
8628
8629
8630
8631
8632
8633
8634
8635
8636
8637
8638
8639
8640
8641
8642
8643
8644
8645
8646
8647
8648
8649
8650
8651
8652
8653
8654
8655
8656
8657
8658
8659
8660
8661
8662
8663
8664
8665
8666
8667
8668
8669
8670
8671
8672
8673
8674
8675
8676
8677
8678
8679
8680
8681
8682
8683
8684
8685
8686
8687
8688
8689
8690
8691
8692
8693
8694
8695
8696
8697
8698
8699
8700
8701
8702
8703
8704
8705
8706
8707
8708
8709
8710
8711
8712
8713
8714
8715
8716
8717
8718
8719
8720
8721
8722
8723
8724
8725
8726
8727
8728
8729
8730
8731
8732
8733
8734
8735
8736
8737
8738
8739
8740
8741
8742
8743
8744
8745
8746
8747
8748
8749
8750
8751
8752
8753
8754
8755
8756
8757
8758
8759
8760
8761
8762
8763
8764
8765
8766
8767
8768
8769
8770
8771
8772
8773
8774
8775
8776
8777
8778
8779
8780
8781
8782
8783
8784
8785
8786
8787
8788
8789
8790
8791
8792
8793
8794
8795
8796
8797
8798
8799
8800
8801
8802
8803
8804
8805
8806
8807
8808
8809
8810
8811
8812
8813
8814
8815
8816
8817
8818
8819
8820
8821
8822
8823
8824
8825
8826
8827
8828
8829
8830
8831
8832
8833
8834
8835
8836
8837
8838
8839
8840
8841
8842
8843
8844
8845
8846
8847
8848
8849
8850
8851
8852
8853
8854
8855
8856
8857
8858
8859
8860
8861
8862
8863
8864
8865
8866
8867
8868
8869
8870
8871
8872
8873
8874
8875
8876
8877
8878
8879
8880
8881
8882
8883
8884
8885
8886
8887
8888
8889
8890
8891
8892
8893
8894
8895
8896
8897
8898
8899
8900
8901
8902
8903
8904
8905
8906
8907
8908
8909
8910
8911
8912
8913
8914
8915
8916
8917
8918
8919
8920
8921
8922
8923
8924
8925
8926
8927
8928
8929
8930
8931
8932
8933
8934
8935
8936
8937
8938
8939
8940
8941
8942
8943
8944
8945
8946
8947
8948
8949
8950
8951
8952
8953
8954
8955
8956
8957
8958
8959
8960
8961
8962
8963
8964
8965
8966
8967
8968
8969
8970
8971
8972
8973
8974
8975
8976
8977
8978
8979
8980
8981
8982
8983
8984
8985
8986
8987
8988
8989
8990
8991
8992
8993
8994
8995
8996
8997
8998
8999
9000
9001
9002
9003
9004
9005
9006
9007
9008
9009
9010
9011
9012
9013
9014
9015
9016
9017
9018
9019
9020
9021
9022
9023
9024
9025
9026
9027
9028
9029
9030
9031
9032
9033
9034
9035
9036
9037
9038
9039
9040
9041
9042
9043
9044
9045
9046
9047
9048
9049
9050
9051
9052
9053
9054
9055
9056
9057
9058
9059
9060
9061
9062
9063
9064
9065
9066
9067
9068
9069
9070
9071
9072
9073
9074
9075
9076
9077
9078
9079
9080
9081
9082
9083
9084
9085
9086
9087
9088
9089
9090
9091
9092
9093
9094
9095
9096
9097
9098
9099
9100
9101
9102
9103
9104
9105
9106
9107
9108
9109
9110
9111
9112
9113
9114
9115
9116
9117
9118
9119
9120
9121
9122
9123
9124
9125
9126
9127
9128
9129
9130
9131
9132
9133
9134
9135
9136
9137
9138
9139
9140
9141
9142
9143
9144
9145
9146
9147
9148
9149
9150
9151
9152
9153
9154
9155
9156
9157
9158
9159
9160
9161
9162
9163
9164
9165
9166
9167
9168
9169
9170
9171
9172
9173
9174
9175
9176
9177
9178
9179
9180
9181
9182
9183
9184
9185
9186
9187
9188
9189
9190
9191
9192
9193
9194
9195
9196
9197
9198
9199
9200
9201
9202
9203
9204
9205
9206
9207
9208
9209
9210
9211
9212
9213
9214
9215
9216
9217
9218
9219
9220
9221
9222
9223
9224
9225
9226
9227
9228
9229
9230
9231
9232
9233
9234
9235
9236
9237
9238
9239
9240
9241
9242
9243
9244
9245
9246
9247
9248
9249
9250
9251
9252
9253
9254
9255
9256
9257
9258
9259
9260
9261
9262
9263
9264
9265
9266
9267
9268
9269
9270
9271
9272
9273
9274
9275
9276
9277
9278
9279
9280
9281
9282
9283
9284
9285
9286
9287
9288
9289
9290
9291
9292
9293
9294
9295
9296
9297
9298
9299
9300
9301
9302
9303
9304
9305
9306
9307
9308
9309
9310
9311
9312
9313
9314
9315
9316
9317
9318
9319
9320
9321
9322
9323
9324
9325
9326
9327
9328
9329
9330
9331
9332
9333
9334
9335
9336
9337
9338
9339
9340
9341
9342
9343
9344
9345
9346
9347
9348
9349
9350
9351
9352
9353
9354
9355
9356
9357
9358
9359
9360
9361
9362
9363
9364
9365
9366
9367
9368
9369
9370
9371
9372
9373
9374
9375
9376
9377
9378
9379
9380
9381
9382
9383
9384
9385
9386
9387
9388
9389
9390
9391
9392
9393
9394
9395
9396
9397
9398
9399
9400
9401
9402
9403
9404
9405
9406
9407
9408
9409
9410
9411
9412
9413
9414
9415
9416
9417
9418
9419
9420
9421
9422
9423
9424
9425
9426
9427
9428
9429
9430
9431
9432
9433
9434
9435
9436
9437
9438
9439
9440
9441
9442
9443
9444
9445
9446
9447
9448
9449
9450
9451
9452
9453
9454
9455
9456
9457
9458
9459
9460
9461
9462
9463
9464
9465
9466
9467
9468
9469
9470
9471
9472
9473
9474
9475
9476
9477
9478
9479
9480
9481
9482
9483
9484
9485
9486
9487
9488
9489
9490
9491
9492
9493
9494
9495
9496
9497
9498
9499
9500
9501
9502
9503
9504
9505
9506
9507
9508
9509
9510
9511
9512
9513
9514
9515
9516
9517
9518
9519
9520
9521
9522
9523
9524
9525
9526
9527
9528
9529
9530
9531
9532
9533
9534
9535
9536
9537
9538
9539
9540
9541
9542
9543
9544
9545
9546
9547
9548
9549
9550
9551
9552
9553
9554
9555
9556
9557
9558
9559
9560
9561
9562
9563
9564
9565
9566
9567
9568
9569
9570
9571
9572
9573
9574
9575
9576
9577
9578
9579
9580
9581
9582
9583
9584
9585
9586
9587
9588
9589
9590
9591
9592
9593
9594
9595
9596
9597
9598
9599
9600
9601
9602
9603
9604
9605
9606
9607
9608
9609
9610
9611
9612
9613
9614
9615
9616
9617
9618
9619
9620
9621
9622
9623
9624
9625
9626
9627
9628
9629
9630
9631
9632
9633
9634
9635
9636
9637
9638
9639
9640
9641
9642
9643
9644
9645
9646
9647
9648
9649
9650
9651
9652
9653
9654
9655
9656
9657
9658
9659
9660
9661
9662
9663
9664
9665
9666
9667
9668
9669
9670
9671
9672
9673
9674
9675
9676
9677
9678
9679
9680
9681
9682
9683
9684
9685
9686
9687
9688
9689
9690
9691
9692
9693
9694
9695
9696
9697
9698
9699
9700
9701
9702
9703
9704
9705
9706
9707
9708
9709
9710
9711
9712
9713
9714
9715
9716
9717
9718
9719
9720
9721
9722
9723
9724
9725
9726
9727
9728
9729
9730
9731
9732
9733
9734
9735
9736
9737
9738
9739
9740
9741
9742
9743
9744
9745
9746
9747
9748
9749
9750
9751
9752
9753
9754
9755
9756
9757
9758
9759
9760
9761
9762
9763
9764
9765
9766
9767
9768
9769
9770
9771
9772
9773
9774
9775
9776
9777
9778
9779
9780
9781
9782
9783
9784
9785
9786
9787
9788
9789
9790
9791
9792
9793
9794
9795
9796
9797
9798
9799
9800
9801
9802
9803
9804
9805
9806
9807
9808
9809
9810
9811
9812
9813
9814
9815
9816
9817
9818
9819
9820
9821
9822
9823
9824
9825
9826
9827
9828
9829
9830
9831
9832
9833
9834
9835
9836
9837
9838
9839
9840
9841
9842
9843
9844
9845
9846
9847
9848
9849
9850
9851
9852
9853
9854
9855
9856
9857
9858
9859
9860
9861
9862
9863
9864
9865
9866
9867
9868
9869
9870
9871
9872
9873
9874
9875
9876
9877
9878
9879
9880
9881
9882
9883
9884
9885
9886
9887
9888
9889
9890
9891
9892
9893
9894
9895
9896
9897
9898
9899
9900
9901
9902
9903
9904
9905
9906
9907
9908
9909
9910
9911
9912
9913
9914
9915
9916
9917
9918
9919
9920
9921
9922
9923
9924
9925
9926
9927
9928
9929
9930
9931
9932
9933
9934
9935
9936
9937
9938
9939
9940
9941
9942
9943
9944
9945
9946
9947
9948
9949
9950
9951
9952
9953
9954
9955
9956
9957
9958
9959
9960
9961
9962
9963
9964
9965
9966
9967
9968
9969
9970
9971
9972
9973
9974
9975
9976
9977
9978
9979
9980
9981
9982
9983
9984
9985
9986
9987
9988
9989
9990
9991
9992
9993
9994
9995
9996
9997
9998
9999
10000
10001
10002
10003
10004
10005
10006
10007
10008
10009
10010
10011
10012
10013
10014
10015
10016
10017
10018
10019
10020
10021
10022
10023
10024
10025
10026
10027
10028
10029
10030
10031
10032
10033
10034
10035
10036
10037
10038
10039
10040
10041
10042
10043
10044
10045
10046
10047
10048
10049
10050
10051
10052
10053
10054
10055
10056
10057
10058
10059
10060
10061
10062
10063
10064
10065
10066
10067
10068
10069
10070
10071
10072
10073
10074
10075
10076
10077
10078
10079
10080
10081
10082
10083
10084
10085
10086
10087
10088
10089
10090
10091
10092
10093
10094
10095
10096
10097
10098
10099
10100
10101
10102
10103
10104
10105
10106
10107
10108
10109
10110
10111
10112
10113
10114
10115
10116
10117
10118
10119
10120
10121
10122
10123
10124
10125
10126
10127
10128
10129
10130
10131
10132
10133
10134
10135
10136
10137
10138
10139
10140
10141
10142
10143
10144
10145
10146
10147
10148
10149
10150
10151
10152
10153
10154
10155
10156
10157
10158
10159
10160
10161
10162
10163
10164
10165
10166
10167
10168
10169
10170
10171
10172
10173
10174
10175
10176
10177
10178
10179
10180
10181
10182
10183
10184
10185
10186
10187
10188
10189
10190
10191
10192
10193
10194
10195
10196
10197
10198
10199
10200
10201
10202
10203
10204
10205
10206
10207
10208
10209
10210
10211
10212
10213
10214
10215
10216
10217
10218
10219
10220
10221
10222
10223
10224
10225
10226
10227
10228
10229
10230
10231
10232
10233
10234
10235
10236
10237
10238
10239
10240
10241
10242
10243
10244
10245
10246
10247
10248
10249
10250
10251
10252
10253
10254
10255
10256
10257
10258
10259
10260
10261
10262
10263
10264
10265
10266
10267
10268
10269
10270
10271
10272
10273
10274
10275
10276
10277
10278
10279
10280
10281
10282
10283
10284
10285
10286
10287
10288
10289
10290
10291
10292
10293
10294
10295
10296
10297
10298
10299
10300
10301
10302
10303
10304
10305
10306
10307
10308
10309
10310
10311
10312
10313
10314
10315
10316
10317
10318
10319
10320
10321
10322
10323
10324
10325
10326
10327
10328
10329
10330
10331
10332
10333
10334
10335
10336
10337
10338
10339
10340
10341
10342
10343
10344
10345
10346
10347
10348
10349
10350
10351
10352
10353
10354
10355
10356
10357
10358
10359
10360
10361
10362
10363
10364
10365
10366
10367
10368
10369
10370
10371
10372
10373
10374
10375
10376
10377
10378
10379
10380
10381
10382
10383
10384
10385
10386
10387
10388
10389
10390
10391
10392
10393
10394
10395
10396
10397
10398
10399
10400
10401
10402
10403
10404
10405
10406
10407
10408
10409
10410
10411
10412
10413
10414
10415
10416
10417
10418
10419
10420
10421
10422
10423
10424
10425
10426
10427
10428
10429
10430
10431
10432
10433
10434
10435
10436
10437
10438
10439
10440
10441
10442
10443
10444
10445
10446
10447
10448
10449
10450
10451
10452
10453
10454
10455
10456
10457
10458
10459
10460
10461
10462
10463
10464
10465
10466
10467
10468
10469
10470
10471
10472
10473
10474
10475
10476
10477
10478
10479
10480
10481
10482
10483
10484
10485
10486
10487
10488
10489
10490
10491
10492
10493
10494
10495
10496
10497
10498
10499
10500
10501
10502
10503
10504
10505
10506
10507
10508
10509
10510
10511
10512
10513
10514
10515
10516
10517
10518
10519
10520
10521
10522
10523
10524
10525
10526
10527
10528
10529
10530
10531
10532
10533
10534
10535
10536
10537
10538
10539
10540
10541
10542
10543
10544
10545
10546
10547
10548
10549
10550
10551
10552
10553
10554
10555
10556
10557
10558
10559
10560
10561
10562
10563
10564
10565
10566
10567
10568
10569
10570
10571
10572
10573
10574
10575
10576
10577
10578
10579
10580
10581
10582
10583
10584
10585
10586
10587
10588
10589
10590
10591
10592
10593
10594
10595
10596
10597
10598
10599
10600
10601
10602
10603
10604
10605
10606
10607
10608
10609
10610
10611
10612
10613
10614
10615
10616
10617
10618
10619
10620
10621
10622
10623
10624
10625
10626
10627
10628
10629
10630
10631
10632
10633
10634
10635
10636
10637
10638
10639
10640
10641
10642
10643
10644
10645
10646
10647
10648
10649
10650
10651
10652
10653
10654
10655
10656
10657
10658
10659
10660
10661
10662
10663
10664
10665
10666
10667
10668
10669
10670
10671
10672
10673
10674
10675
10676
10677
10678
10679
10680
10681
10682
10683
10684
10685
10686
10687
10688
10689
10690
10691
10692
10693
10694
10695
10696
10697
10698
10699
10700
10701
10702
10703
10704
10705
10706
10707
10708
10709
10710
10711
10712
10713
10714
10715
10716
10717
10718
10719
10720
10721
10722
10723
10724
10725
10726
10727
10728
10729
10730
10731
10732
10733
10734
10735
10736
10737
10738
10739
10740
10741
10742
10743
10744
10745
10746
10747
10748
10749
10750
10751
10752
10753
10754
10755
10756
10757
10758
10759
10760
10761
10762
10763
10764
10765
10766
10767
10768
10769
10770
10771
10772
10773
10774
10775
10776
10777
10778
10779
10780
10781
10782
10783
10784
10785
10786
10787
10788
10789
10790
10791
10792
10793
10794
10795
10796
10797
10798
10799
10800
10801
10802
10803
10804
10805
10806
10807
10808
10809
10810
10811
10812
10813
10814
10815
10816
10817
10818
10819
10820
10821
10822
10823
10824
10825
10826
10827
10828
10829
10830
10831
10832
10833
10834
10835
10836
10837
10838
10839
10840
10841
10842
10843
10844
10845
10846
10847
10848
10849
10850
10851
10852
10853
10854
10855
10856
10857
10858
10859
10860
10861
10862
10863
10864
10865
10866
10867
10868
10869
10870
10871
10872
10873
10874
10875
10876
10877
10878
10879
10880
10881
10882
10883
10884
10885
10886
10887
10888
10889
10890
10891
10892
10893
10894
10895
10896
10897
10898
10899
10900
10901
10902
10903
10904
10905
10906
10907
10908
10909
10910
10911
10912
10913
10914
10915
10916
10917
10918
10919
10920
10921
10922
10923
10924
10925
10926
10927
10928
10929
10930
10931
10932
10933
10934
10935
10936
10937
10938
10939
10940
10941
10942
10943
10944
10945
10946
10947
10948
10949
10950
10951
10952
10953
10954
10955
10956
10957
10958
10959
10960
10961
10962
10963
10964
10965
10966
10967
10968
10969
10970
10971
10972
10973
10974
10975
10976
10977
10978
10979
10980
10981
10982
10983
10984
10985
10986
10987
10988
10989
10990
10991
10992
10993
10994
10995
10996
10997
10998
10999
11000
11001
11002
11003
11004
11005
11006
11007
11008
11009
11010
11011
11012
11013
11014
11015
11016
11017
11018
11019
11020
11021
11022
11023
11024
11025
11026
11027
11028
11029
11030
11031
11032
11033
11034
11035
11036
11037
11038
11039
11040
11041
11042
11043
11044
11045
11046
11047
11048
11049
11050
11051
11052
11053
11054
11055
11056
11057
11058
11059
11060
11061
11062
11063
11064
11065
11066
11067
11068
11069
11070
11071
11072
11073
11074
11075
11076
11077
11078
11079
11080
11081
11082
11083
11084
11085
11086
11087
11088
11089
11090
11091
11092
11093
11094
11095
11096
11097
11098
11099
11100
11101
11102
11103
11104
11105
11106
11107
11108
11109
11110
11111
11112
11113
11114
11115
11116
11117
11118
11119
11120
11121
11122
11123
11124
11125
11126
11127
11128
11129
11130
11131
11132
11133
11134
11135
11136
11137
11138
11139
11140
11141
11142
11143
11144
11145
11146
11147
11148
11149
11150
11151
11152
11153
11154
11155
11156
11157
11158
11159
11160
11161
11162
11163
11164
11165
11166
11167
11168
11169
11170
11171
11172
11173
11174
11175
11176
11177
11178
11179
11180
11181
11182
11183
11184
11185
11186
11187
11188
11189
11190
11191
11192
11193
11194
11195
11196
11197
11198
11199
11200
11201
11202
11203
11204
11205
11206
11207
11208
11209
11210
11211
11212
11213
11214
11215
11216
11217
11218
11219
11220
11221
11222
11223
11224
11225
11226
11227
11228
11229
11230
11231
11232
11233
11234
11235
11236
11237
11238
11239
11240
11241
11242
11243
11244
11245
11246
11247
11248
11249
11250
11251
11252
11253
11254
11255
11256
11257
11258
11259
11260
11261
11262
11263
11264
11265
11266
11267
11268
11269
11270
11271
11272
11273
11274
11275
11276
11277
11278
11279
11280
11281
11282
11283
11284
11285
11286
11287
11288
11289
11290
11291
11292
11293
11294
11295
11296
11297
11298
11299
11300
11301
11302
11303
11304
11305
11306
11307
11308
11309
11310
11311
11312
11313
11314
11315
11316
11317
11318
11319
11320
11321
11322
11323
11324
11325
11326
11327
11328
11329
11330
11331
11332
11333
11334
11335
11336
11337
11338
11339
11340
11341
11342
11343
11344
11345
11346
11347
11348
11349
11350
11351
11352
11353
11354
11355
11356
11357
11358
11359
11360
11361
11362
11363
11364
11365
11366
11367
11368
11369
11370
11371
11372
11373
11374
11375
11376
11377
11378
11379
11380
11381
11382
11383
11384
11385
11386
11387
11388
11389
11390
11391
11392
11393
11394
11395
11396
11397
11398
11399
11400
11401
11402
11403
11404
11405
11406
11407
11408
11409
11410
11411
11412
11413
11414
11415
11416
11417
11418
11419
11420
11421
11422
11423
11424
11425
11426
11427
11428
11429
11430
11431
11432
11433
11434
11435
11436
11437
11438
11439
11440
11441
11442
11443
11444
11445
11446
11447
11448
11449
11450
11451
11452
11453
11454
11455
11456
11457
11458
11459
11460
11461
11462
11463
11464
11465
11466
11467
11468
11469
11470
11471
11472
11473
11474
11475
11476
11477
11478
11479
11480
11481
11482
11483
11484
11485
11486
11487
11488
11489
11490
11491
11492
11493
11494
11495
11496
11497
11498
11499
11500
11501
11502
11503
11504
11505
11506
11507
11508
11509
11510
11511
11512
11513
11514
11515
11516
11517
11518
11519
11520
11521
11522
11523
11524
11525
11526
11527
11528
11529
11530
11531
11532
11533
11534
11535
11536
11537
11538
11539
11540
11541
11542
11543
11544
11545
11546
11547
11548
11549
11550
11551
11552
11553
11554
11555
11556
11557
11558
11559
11560
11561
11562
11563
11564
11565
11566
11567
11568
11569
11570
11571
11572
11573
11574
11575
11576
11577
11578
11579
11580
11581
11582
11583
11584
11585
11586
11587
11588
11589
11590
11591
11592
11593
11594
11595
11596
11597
11598
11599
11600
11601
11602
11603
11604
11605
11606
11607
11608
11609
11610
11611
11612
11613
11614
11615
11616
11617
11618
11619
11620
11621
11622
11623
11624
11625
11626
11627
11628
11629
11630
11631
11632
11633
11634
11635
11636
11637
11638
11639
11640
11641
11642
11643
11644
11645
11646
11647
11648
11649
11650
11651
11652
11653
11654
11655
11656
11657
11658
11659
11660
11661
11662
11663
11664
11665
11666
11667
11668
11669
11670
11671
11672
11673
11674
11675
11676
11677
11678
11679
11680
11681
11682
11683
11684
11685
11686
11687
11688
11689
11690
11691
11692
11693
11694
11695
11696
11697
11698
11699
11700
11701
11702
11703
11704
11705
11706
11707
11708
11709
11710
11711
11712
11713
11714
11715
11716
11717
11718
11719
11720
11721
11722
11723
11724
11725
11726
11727
11728
11729
11730
11731
11732
11733
11734
11735
11736
11737
11738
11739
11740
11741
11742
11743
11744
11745
11746
11747
11748
11749
11750
11751
11752
11753
11754
11755
11756
11757
11758
11759
11760
11761
11762
11763
11764
11765
11766
11767
11768
11769
11770
11771
11772
11773
11774
11775
11776
11777
11778
11779
11780
11781
11782
11783
11784
11785
11786
11787
11788
11789
11790
11791
11792
11793
11794
11795
11796
11797
11798
11799
11800
11801
11802
11803
11804
11805
11806
11807
11808
11809
11810
11811
11812
11813
11814
11815
11816
11817
11818
11819
11820
11821
11822
11823
11824
11825
11826
11827
11828
11829
11830
11831
11832
11833
11834
11835
11836
11837
11838
11839
11840
11841
11842
11843
11844
11845
11846
11847
11848
11849
11850
11851
11852
11853
11854
11855
11856
11857
11858
11859
11860
11861
11862
11863
11864
11865
11866
11867
11868
11869
11870
11871
11872
11873
11874
11875
11876
11877
11878
11879
11880
11881
11882
11883
11884
11885
11886
11887
11888
11889
11890
11891
11892
11893
11894
11895
11896
11897
11898
11899
11900
11901
11902
11903
11904
11905
11906
11907
11908
11909
11910
11911
11912
11913
11914
11915
11916
11917
11918
11919
11920
11921
11922
11923
11924
11925
11926
11927
11928
11929
11930
11931
11932
11933
11934
11935
11936
11937
11938
11939
11940
11941
11942
11943
11944
11945
11946
11947
11948
11949
11950
11951
11952
11953
11954
11955
11956
11957
11958
11959
11960
11961
11962
11963
11964
11965
11966
11967
11968
11969
11970
11971
11972
11973
11974
11975
11976
11977
11978
11979
11980
11981
11982
11983
11984
11985
11986
11987
11988
11989
11990
11991
11992
11993
11994
11995
11996
11997
11998
11999
12000
12001
12002
12003
12004
12005
12006
12007
12008
12009
12010
12011
12012
12013
12014
12015
12016
12017
12018
12019
12020
12021
12022
12023
12024
12025
12026
12027
12028
12029
12030
12031
12032
12033
12034
12035
12036
12037
12038
12039
12040
12041
12042
12043
12044
12045
12046
12047
12048
12049
12050
12051
12052
12053
12054
12055
12056
12057
12058
12059
12060
12061
12062
12063
12064
12065
12066
12067
12068
12069
12070
12071
12072
12073
12074
12075
12076
12077
12078
12079
12080
12081
12082
12083
12084
12085
12086
12087
12088
12089
12090
12091
12092
12093
12094
12095
12096
12097
12098
12099
12100
12101
12102
12103
12104
12105
12106
12107
12108
12109
12110
12111
12112
12113
12114
12115
12116
12117
12118
12119
12120
12121
12122
12123
12124
12125
12126
12127
12128
12129
12130
12131
12132
12133
12134
12135
12136
12137
12138
12139
12140
12141
12142
12143
12144
12145
12146
12147
12148
12149
12150
12151
12152
12153
12154
12155
12156
12157
12158
12159
12160
12161
12162
12163
12164
12165
12166
12167
12168
12169
12170
12171
12172
12173
12174
12175
12176
12177
12178
12179
12180
12181
12182
12183
12184
12185
12186
12187
12188
12189
12190
12191
12192
12193
12194
12195
12196
12197
12198
12199
12200
12201
12202
12203
12204
12205
12206
12207
12208
12209
12210
12211
12212
12213
12214
12215
12216
12217
12218
12219
12220
12221
12222
12223
12224
12225
12226
12227
12228
12229
12230
12231
12232
12233
12234
12235
12236
12237
12238
12239
12240
12241
12242
12243
12244
12245
12246
12247
12248
12249
12250
12251
12252
12253
12254
12255
12256
12257
12258
12259
12260
12261
12262
12263
12264
12265
12266
12267
12268
12269
12270
12271
12272
12273
12274
12275
12276
12277
12278
12279
12280
12281
12282
12283
12284
12285
12286
12287
12288
12289
12290
12291
12292
12293
12294
12295
12296
12297
12298
12299
12300
12301
12302
12303
12304
12305
12306
12307
12308
12309
12310
12311
12312
12313
12314
12315
12316
12317
12318
12319
12320
12321
12322
12323
12324
12325
12326
12327
12328
12329
12330
12331
12332
12333
12334
12335
12336
12337
12338
12339
12340
12341
12342
12343
12344
12345
12346
12347
12348
12349
12350
12351
12352
12353
12354
12355
12356
12357
12358
12359
12360
12361
12362
12363
12364
12365
12366
12367
12368
12369
12370
12371
12372
12373
12374
12375
12376
12377
12378
12379
12380
12381
12382
12383
12384
12385
12386
12387
12388
12389
12390
12391
12392
12393
12394
12395
12396
12397
12398
12399
12400
12401
12402
12403
12404
12405
12406
12407
12408
12409
12410
12411
12412
12413
12414
12415
12416
12417
12418
12419
12420
12421
12422
12423
12424
12425
12426
12427
12428
12429
12430
12431
12432
12433
12434
12435
12436
12437
12438
12439
12440
12441
12442
12443
12444
12445
12446
12447
12448
12449
12450
12451
12452
12453
12454
12455
12456
12457
12458
12459
12460
12461
12462
12463
12464
12465
12466
12467
12468
12469
12470
12471
12472
12473
12474
12475
12476
12477
12478
12479
12480
12481
12482
12483
12484
12485
12486
12487
12488
12489
12490
12491
12492
12493
12494
12495
12496
12497
12498
12499
12500
12501
12502
12503
12504
12505
12506
12507
12508
12509
12510
12511
12512
12513
12514
12515
12516
12517
12518
12519
12520
12521
12522
12523
12524
12525
12526
12527
12528
12529
12530
12531
12532
12533
12534
12535
12536
12537
12538
12539
12540
12541
12542
12543
12544
12545
12546
12547
12548
12549
12550
12551
12552
12553
12554
12555
12556
12557
12558
12559
12560
12561
12562
12563
12564
12565
12566
12567
12568
12569
12570
12571
12572
12573
12574
12575
12576
12577
12578
12579
12580
12581
12582
12583
12584
12585
12586
12587
12588
12589
12590
12591
12592
12593
12594
12595
12596
12597
12598
12599
12600
12601
12602
12603
12604
12605
12606
12607
12608
12609
12610
12611
12612
12613
12614
12615
12616
12617
12618
12619
12620
12621
12622
12623
12624
12625
12626
12627
12628
12629
12630
12631
12632
12633
12634
12635
12636
12637
12638
12639
12640
12641
12642
12643
12644
12645
12646
12647
12648
12649
12650
12651
12652
12653
12654
12655
12656
12657
12658
12659
12660
12661
12662
12663
12664
12665
12666
12667
12668
12669
12670
12671
12672
12673
12674
12675
12676
12677
12678
12679
12680
12681
12682
12683
12684
12685
12686
12687
12688
12689
12690
12691
12692
12693
12694
12695
12696
12697
12698
12699
12700
12701
12702
12703
12704
12705
12706
12707
12708
12709
12710
12711
12712
12713
12714
12715
12716
12717
12718
12719
12720
12721
12722
12723
12724
12725
12726
12727
12728
12729
12730
12731
12732
12733
12734
12735
12736
12737
12738
12739
12740
12741
12742
12743
12744
12745
12746
12747
12748
12749
12750
12751
12752
12753
12754
12755
12756
12757
12758
12759
12760
12761
12762
12763
12764
12765
12766
12767
12768
12769
12770
12771
12772
12773
12774
12775
12776
12777
12778
12779
12780
12781
12782
12783
12784
12785
12786
12787
12788
12789
12790
12791
12792
12793
12794
12795
12796
12797
12798
12799
12800
12801
12802
12803
12804
12805
12806
12807
12808
12809
12810
12811
12812
12813
12814
12815
12816
12817
12818
12819
12820
12821
12822
12823
12824
12825
12826
12827
12828
12829
12830
12831
12832
12833
12834
12835
12836
12837
12838
12839
12840
12841
12842
12843
12844
12845
12846
12847
12848
12849
12850
12851
12852
12853
12854
12855
12856
12857
12858
12859
12860
12861
12862
12863
12864
12865
12866
12867
12868
12869
12870
12871
12872
12873
12874
12875
12876
12877
12878
12879
12880
12881
12882
12883
12884
12885
12886
12887
12888
12889
12890
12891
12892
12893
12894
12895
12896
12897
12898
12899
12900
12901
12902
12903
12904
12905
12906
12907
12908
12909
12910
12911
12912
12913
12914
12915
12916
12917
12918
12919
12920
12921
12922
12923
12924
12925
12926
12927
12928
12929
12930
12931
12932
12933
12934
12935
12936
12937
12938
12939
12940
12941
12942
12943
12944
12945
12946
12947
12948
12949
12950
12951
12952
12953
12954
12955
12956
12957
12958
12959
12960
12961
12962
12963
12964
12965
12966
12967
12968
12969
12970
12971
12972
12973
12974
12975
12976
12977
12978
12979
12980
12981
12982
12983
12984
12985
12986
12987
12988
12989
12990
12991
12992
12993
12994
12995
12996
12997
12998
12999
13000
13001
13002
13003
13004
13005
13006
13007
13008
13009
13010
13011
13012
13013
13014
13015
13016
13017
13018
13019
13020
13021
13022
13023
13024
13025
13026
13027
13028
13029
13030
13031
13032
13033
13034
13035
13036
13037
13038
13039
13040
13041
13042
13043
13044
13045
13046
13047
13048
13049
13050
13051
13052
13053
13054
13055
13056
13057
13058
13059
13060
13061
13062
13063
13064
13065
13066
13067
13068
13069
13070
13071
13072
13073
13074
13075
13076
13077
13078
13079
13080
13081
13082
13083
13084
13085
13086
13087
13088
13089
13090
13091
13092
13093
13094
13095
13096
13097
13098
13099
13100
13101
13102
13103
13104
13105
13106
13107
13108
13109
13110
13111
13112
13113
13114
13115
13116
13117
13118
13119
13120
13121
13122
13123
13124
13125
13126
13127
13128
13129
13130
13131
13132
13133
13134
13135
13136
13137
13138
13139
13140
13141
13142
13143
13144
13145
13146
13147
13148
13149
13150
13151
13152
13153
13154
13155
13156
13157
13158
13159
13160
13161
13162
13163
13164
13165
13166
13167
13168
13169
13170
13171
13172
13173
13174
13175
13176
13177
13178
13179
13180
13181
13182
13183
13184
13185
13186
13187
13188
13189
13190
13191
13192
13193
13194
13195
13196
13197
13198
13199
13200
13201
13202
13203
13204
13205
13206
13207
13208
13209
13210
13211
13212
13213
13214
13215
13216
13217
13218
13219
13220
13221
13222
13223
13224
13225
13226
13227
13228
13229
13230
13231
13232
13233
13234
13235
13236
13237
13238
13239
13240
13241
13242
13243
13244
13245
13246
13247
13248
13249
13250
13251
13252
13253
13254
13255
13256
13257
13258
13259
13260
13261
13262
13263
13264
13265
13266
13267
13268
13269
13270
13271
13272
13273
13274
13275
13276
13277
13278
13279
13280
13281
13282
13283
13284
13285
13286
13287
13288
13289
13290
13291
13292
13293
13294
13295
13296
13297
13298
13299
13300
13301
13302
13303
13304
13305
13306
13307
13308
13309
13310
13311
13312
13313
13314
13315
13316
13317
13318
13319
13320
13321
13322
13323
13324
13325
13326
13327
13328
13329
13330
13331
13332
13333
13334
13335
13336
13337
13338
13339
13340
13341
13342
13343
13344
13345
13346
13347
13348
13349
13350
13351
13352
13353
13354
13355
13356
13357
13358
13359
13360
13361
13362
13363
13364
13365
13366
13367
13368
13369
13370
13371
13372
13373
13374
13375
13376
13377
13378
13379
13380
13381
13382
13383
13384
13385
13386
13387
13388
13389
13390
13391
13392
13393
13394
13395
13396
13397
13398
13399
13400
13401
13402
13403
13404
13405
13406
13407
13408
13409
13410
13411
13412
13413
13414
13415
13416
13417
13418
13419
13420
13421
13422
13423
13424
13425
13426
13427
13428
13429
13430
13431
13432
13433
13434
13435
13436
13437
13438
13439
13440
13441
13442
13443
13444
13445
13446
13447
13448
13449
13450
13451
13452
13453
13454
13455
13456
13457
13458
13459
13460
13461
13462
13463
13464
13465
13466
13467
13468
13469
13470
13471
13472
13473
13474
13475
13476
13477
13478
13479
13480
13481
13482
13483
13484
13485
13486
13487
13488
13489
13490
13491
13492
13493
13494
13495
13496
13497
13498
13499
13500
13501
13502
13503
13504
13505
13506
13507
13508
13509
13510
13511
13512
13513
13514
13515
13516
13517
13518
13519
13520
13521
13522
13523
13524
13525
13526
13527
13528
13529
13530
13531
13532
13533
13534
13535
13536
13537
13538
13539
13540
13541
13542
13543
13544
13545
13546
13547
13548
13549
13550
13551
13552
13553
13554
13555
13556
13557
13558
13559
13560
13561
13562
13563
13564
13565
13566
13567
13568
13569
13570
13571
13572
13573
13574
13575
13576
13577
13578
13579
13580
13581
13582
13583
13584
13585
13586
13587
13588
13589
13590
13591
13592
13593
13594
13595
13596
13597
13598
13599
13600
13601
13602
13603
13604
13605
13606
13607
13608
13609
13610
13611
13612
13613
13614
13615
13616
13617
13618
13619
13620
13621
13622
13623
13624
13625
13626
13627
13628
13629
13630
13631
13632
13633
13634
13635
13636
13637
13638
13639
13640
13641
13642
13643
13644
13645
13646
13647
13648
13649
13650
13651
13652
13653
13654
13655
13656
13657
13658
13659
13660
13661
13662
13663
13664
13665
13666
13667
13668
13669
13670
13671
13672
13673
13674
13675
13676
13677
13678
13679
13680
13681
13682
13683
13684
13685
13686
13687
13688
13689
13690
13691
13692
13693
13694
13695
13696
13697
13698
13699
13700
13701
13702
13703
13704
13705
13706
13707
13708
13709
13710
13711
13712
13713
13714
13715
13716
13717
13718
13719
13720
13721
13722
13723
13724
13725
13726
13727
13728
13729
13730
13731
13732
13733
13734
13735
13736
13737
13738
13739
13740
13741
13742
13743
13744
13745
13746
13747
13748
13749
13750
13751
13752
13753
13754
13755
13756
13757
13758
13759
13760
13761
13762
13763
13764
13765
13766
13767
13768
13769
13770
13771
13772
13773
13774
13775
13776
13777
13778
13779
13780
13781
13782
13783
13784
13785
13786
13787
13788
13789
13790
13791
13792
13793
13794
13795
13796
13797
13798
13799
13800
13801
13802
13803
13804
13805
13806
13807
13808
13809
13810
13811
13812
13813
13814
13815
13816
13817
13818
13819
13820
13821
13822
13823
13824
13825
13826
13827
13828
13829
13830
13831
13832
13833
13834
13835
13836
13837
13838
13839
13840
13841
13842
13843
13844
13845
13846
13847
13848
13849
13850
13851
13852
13853
13854
13855
13856
13857
13858
13859
13860
13861
13862
13863
13864
13865
13866
13867
13868
13869
13870
13871
13872
13873
13874
13875
13876
13877
13878
13879
13880
13881
13882
13883
13884
13885
13886
13887
13888
13889
13890
13891
13892
13893
13894
13895
13896
13897
13898
13899
13900
13901
13902
13903
13904
13905
13906
13907
13908
13909
13910
13911
13912
13913
13914
13915
13916
13917
13918
13919
13920
13921
13922
13923
13924
13925
13926
13927
13928
13929
13930
13931
13932
13933
13934
13935
13936
13937
13938
13939
13940
13941
13942
13943
13944
13945
13946
13947
13948
13949
13950
13951
13952
13953
13954
13955
13956
13957
13958
13959
13960
13961
13962
13963
13964
13965
13966
13967
13968
13969
13970
13971
13972
13973
13974
13975
13976
13977
13978
13979
13980
13981
13982
13983
13984
13985
13986
13987
13988
13989
13990
13991
13992
13993
13994
13995
13996
13997
13998
13999
14000
14001
14002
14003
14004
14005
14006
14007
14008
14009
14010
14011
14012
14013
14014
14015
14016
14017
14018
14019
14020
14021
14022
14023
14024
14025
14026
14027
14028
14029
14030
14031
14032
14033
14034
14035
14036
14037
14038
14039
14040
14041
14042
14043
14044
14045
14046
14047
14048
14049
14050
14051
14052
14053
14054
14055
14056
14057
14058
14059
14060
14061
14062
14063
14064
14065
14066
14067
14068
14069
14070
14071
14072
14073
14074
14075
14076
14077
14078
14079
14080
14081
14082
14083
14084
14085
14086
14087
14088
14089
14090
14091
14092
14093
14094
14095
14096
14097
14098
14099
14100
14101
14102
14103
14104
14105
14106
14107
14108
14109
14110
14111
14112
14113
14114
14115
14116
14117
14118
14119
14120
14121
14122
14123
14124
14125
14126
14127
14128
14129
14130
14131
14132
14133
14134
14135
14136
14137
14138
14139
14140
14141
14142
14143
14144
14145
14146
14147
14148
14149
14150
14151
14152
14153
14154
14155
14156
14157
14158
14159
14160
14161
14162
14163
14164
14165
14166
14167
14168
14169
14170
14171
14172
14173
14174
14175
14176
14177
14178
14179
14180
14181
14182
14183
14184
14185
14186
14187
14188
14189
14190
14191
14192
14193
14194
14195
14196
14197
14198
14199
14200
14201
14202
14203
14204
14205
14206
14207
14208
14209
14210
14211
14212
14213
14214
14215
14216
14217
14218
14219
14220
14221
14222
14223
14224
14225
14226
14227
14228
14229
14230
14231
14232
14233
14234
14235
14236
14237
14238
14239
14240
14241
14242
14243
14244
14245
14246
14247
14248
14249
14250
14251
14252
14253
14254
14255
14256
14257
14258
14259
14260
14261
14262
14263
14264
14265
14266
14267
14268
14269
14270
14271
14272
14273
14274
14275
14276
14277
14278
14279
14280
14281
14282
14283
14284
14285
14286
14287
14288
14289
14290
14291
14292
14293
14294
14295
14296
14297
14298
14299
14300
14301
14302
14303
14304
14305
14306
14307
14308
14309
14310
14311
14312
14313
14314
14315
14316
14317
14318
14319
14320
14321
14322
14323
14324
14325
14326
14327
14328
14329
14330
14331
14332
14333
14334
14335
14336
14337
14338
14339
14340
14341
14342
14343
14344
14345
14346
14347
14348
14349
14350
14351
14352
14353
14354
14355
14356
14357
14358
14359
14360
14361
14362
14363
14364
14365
14366
14367
14368
14369
14370
14371
14372
14373
14374
14375
14376
14377
14378
14379
14380
14381
14382
14383
14384
14385
14386
14387
14388
14389
14390
14391
14392
14393
14394
14395
14396
14397
14398
14399
14400
14401
14402
14403
14404
14405
14406
14407
14408
14409
14410
14411
14412
14413
14414
14415
14416
14417
14418
14419
14420
14421
14422
14423
14424
14425
14426
14427
14428
14429
14430
14431
14432
14433
14434
14435
14436
14437
14438
14439
14440
14441
14442
14443
14444
14445
14446
14447
14448
14449
14450
14451
14452
14453
14454
14455
14456
14457
14458
14459
14460
14461
14462
14463
14464
14465
14466
14467
14468
14469
14470
14471
14472
14473
14474
14475
14476
14477
14478
14479
14480
14481
14482
14483
14484
14485
14486
14487
14488
14489
14490
14491
14492
14493
14494
14495
14496
14497
14498
14499
14500
14501
14502
14503
14504
14505
14506
14507
14508
14509
14510
14511
14512
14513
14514
14515
14516
14517
14518
14519
14520
14521
14522
14523
14524
14525
14526
14527
14528
14529
14530
14531
14532
14533
14534
14535
14536
14537
14538
14539
14540
14541
14542
14543
14544
14545
14546
14547
14548
14549
14550
14551
14552
14553
14554
14555
14556
14557
14558
14559
14560
14561
14562
14563
14564
14565
14566
14567
14568
14569
14570
14571
14572
14573
14574
14575
14576
14577
14578
14579
14580
14581
14582
14583
14584
14585
14586
14587
14588
14589
14590
14591
14592
14593
14594
14595
14596
14597
14598
14599
14600
14601
14602
14603
14604
14605
14606
14607
14608
14609
14610
14611
14612
14613
14614
14615
14616
14617
14618
14619
14620
14621
14622
14623
14624
14625
14626
14627
14628
14629
14630
14631
14632
14633
14634
14635
14636
14637
14638
14639
14640
14641
14642
14643
14644
14645
14646
14647
14648
14649
14650
14651
14652
14653
14654
14655
14656
14657
14658
14659
14660
14661
14662
14663
14664
14665
14666
14667
14668
14669
14670
14671
14672
14673
14674
14675
14676
14677
14678
14679
14680
14681
14682
14683
14684
14685
14686
14687
14688
14689
14690
14691
14692
14693
14694
14695
14696
14697
14698
14699
14700
14701
14702
14703
14704
14705
14706
14707
14708
14709
14710
14711
14712
14713
14714
14715
14716
14717
14718
14719
14720
14721
14722
14723
14724
14725
14726
14727
14728
14729
14730
14731
14732
14733
14734
14735
14736
14737
14738
14739
14740
14741
14742
14743
14744
14745
14746
14747
14748
14749
14750
14751
14752
14753
14754
14755
14756
14757
14758
14759
14760
14761
14762
14763
14764
14765
14766
14767
14768
14769
14770
14771
14772
14773
14774
14775
14776
14777
14778
14779
14780
14781
14782
14783
14784
14785
14786
14787
14788
14789
14790
14791
14792
14793
14794
14795
14796
14797
14798
14799
14800
14801
14802
14803
14804
14805
14806
14807
14808
14809
14810
14811
14812
14813
14814
14815
14816
14817
14818
14819
14820
14821
14822
14823
14824
14825
14826
14827
14828
14829
14830
14831
14832
14833
14834
14835
14836
14837
14838
14839
14840
14841
14842
14843
14844
14845
14846
14847
14848
14849
14850
14851
14852
14853
14854
14855
14856
14857
14858
14859
14860
14861
14862
14863
14864
14865
14866
14867
14868
14869
14870
14871
14872
14873
14874
14875
14876
14877
14878
14879
14880
14881
14882
14883
14884
14885
14886
14887
14888
14889
14890
14891
14892
14893
14894
14895
14896
14897
14898
14899
14900
14901
14902
14903
14904
14905
14906
14907
14908
14909
14910
14911
14912
14913
14914
14915
14916
14917
14918
14919
14920
14921
14922
14923
14924
14925
14926
14927
14928
14929
14930
14931
14932
14933
14934
14935
14936
14937
14938
14939
14940
14941
14942
14943
14944
14945
14946
14947
14948
14949
14950
14951
14952
14953
14954
14955
14956
14957
14958
14959
14960
14961
14962
14963
14964
14965
14966
14967
14968
14969
14970
14971
14972
14973
14974
14975
14976
14977
14978
14979
14980
14981
14982
14983
14984
14985
14986
14987
14988
14989
14990
14991
14992
14993
14994
14995
14996
14997
14998
14999
15000
15001
15002
15003
15004
15005
15006
15007
15008
15009
15010
15011
15012
15013
15014
15015
15016
15017
15018
15019
15020
15021
15022
15023
15024
15025
15026
15027
15028
15029
15030
15031
15032
15033
15034
15035
15036
15037
15038
15039
15040
15041
15042
15043
15044
15045
15046
15047
15048
15049
15050
15051
15052
15053
15054
15055
15056
15057
15058
15059
15060
15061
15062
15063
15064
15065
15066
15067
15068
15069
15070
15071
15072
15073
15074
15075
15076
15077
15078
15079
15080
15081
15082
15083
15084
15085
15086
15087
15088
15089
15090
15091
15092
15093
15094
15095
15096
15097
15098
15099
15100
15101
15102
15103
15104
15105
15106
15107
15108
15109
15110
15111
15112
15113
15114
15115
15116
15117
15118
15119
15120
15121
15122
15123
15124
15125
15126
15127
15128
15129
15130
15131
15132
15133
15134
15135
15136
15137
15138
15139
15140
15141
15142
15143
15144
15145
15146
15147
15148
15149
15150
15151
15152
15153
15154
15155
15156
15157
15158
15159
15160
15161
15162
15163
15164
15165
15166
15167
15168
15169
15170
15171
15172
15173
15174
15175
15176
15177
15178
15179
15180
15181
15182
15183
15184
15185
15186
15187
15188
15189
15190
15191
15192
15193
15194
15195
15196
15197
15198
15199
15200
15201
15202
15203
15204
15205
15206
15207
15208
15209
15210
15211
15212
15213
15214
15215
15216
15217
15218
15219
15220
15221
15222
15223
15224
15225
15226
15227
15228
15229
15230
15231
15232
15233
15234
15235
15236
15237
15238
15239
15240
15241
15242
15243
15244
15245
15246
15247
15248
15249
15250
15251
15252
15253
15254
15255
15256
15257
15258
15259
15260
15261
15262
15263
15264
15265
15266
15267
15268
15269
15270
15271
15272
15273
15274
15275
15276
15277
15278
15279
15280
15281
15282
15283
15284
15285
15286
15287
15288
15289
15290
15291
15292
15293
15294
15295
15296
15297
15298
15299
15300
15301
15302
15303
15304
15305
15306
15307
15308
15309
15310
15311
15312
15313
15314
15315
15316
15317
15318
15319
15320
15321
15322
15323
15324
15325
15326
15327
15328
15329
15330
15331
15332
15333
15334
15335
15336
15337
15338
15339
15340
15341
15342
15343
15344
15345
15346
15347
15348
15349
15350
15351
15352
15353
15354
15355
15356
15357
15358
15359
15360
15361
15362
15363
15364
15365
15366
15367
15368
15369
15370
15371
15372
15373
15374
15375
15376
15377
15378
15379
15380
15381
15382
15383
15384
15385
15386
15387
15388
15389
15390
15391
15392
15393
15394
15395
15396
15397
15398
15399
15400
15401
15402
15403
15404
15405
15406
15407
15408
15409
15410
15411
15412
15413
15414
15415
15416
15417
15418
15419
15420
15421
15422
15423
15424
15425
15426
15427
15428
15429
15430
15431
15432
15433
15434
15435
15436
15437
15438
15439
15440
15441
15442
15443
15444
15445
15446
15447
15448
15449
15450
15451
15452
15453
15454
15455
15456
15457
15458
15459
15460
15461
15462
15463
15464
15465
15466
15467
15468
15469
15470
15471
15472
15473
15474
15475
15476
15477
15478
15479
15480
15481
15482
15483
15484
15485
15486
15487
15488
15489
15490
15491
15492
15493
15494
15495
15496
15497
15498
15499
15500
15501
15502
15503
15504
15505
15506
15507
15508
15509
15510
15511
15512
15513
15514
15515
15516
15517
15518
15519
15520
15521
15522
15523
15524
15525
15526
15527
15528
15529
15530
15531
15532
15533
15534
15535
15536
15537
15538
15539
15540
15541
15542
15543
15544
15545
15546
15547
15548
15549
15550
15551
15552
15553
15554
15555
15556
15557
15558
15559
15560
15561
15562
15563
15564
15565
15566
15567
15568
15569
15570
15571
15572
15573
15574
15575
15576
15577
15578
15579
15580
15581
15582
15583
15584
15585
15586
15587
15588
15589
15590
15591
15592
15593
15594
15595
15596
15597
15598
15599
15600
15601
15602
15603
15604
15605
15606
15607
15608
15609
15610
15611
15612
15613
15614
15615
15616
15617
15618
15619
15620
15621
15622
15623
15624
15625
15626
15627
15628
15629
15630
15631
15632
15633
15634
15635
15636
15637
15638
15639
15640
15641
15642
15643
15644
15645
15646
15647
15648
15649
15650
15651
15652
15653
15654
15655
15656
15657
15658
15659
15660
15661
15662
15663
15664
15665
15666
15667
15668
15669
15670
15671
15672
15673
15674
15675
15676
15677
15678
15679
15680
15681
15682
15683
15684
15685
15686
15687
15688
15689
15690
15691
15692
15693
15694
15695
15696
15697
15698
15699
15700
15701
15702
15703
15704
15705
15706
15707
15708
15709
15710
15711
15712
15713
15714
15715
15716
15717
15718
15719
15720
15721
15722
15723
15724
15725
15726
15727
15728
15729
15730
15731
15732
15733
15734
15735
15736
15737
15738
15739
15740
15741
15742
15743
15744
15745
15746
15747
15748
15749
15750
15751
15752
15753
15754
15755
15756
15757
15758
15759
15760
15761
15762
15763
15764
15765
15766
15767
15768
15769
15770
15771
15772
15773
15774
15775
15776
15777
15778
15779
15780
15781
15782
15783
15784
15785
15786
15787
15788
15789
15790
15791
15792
15793
15794
15795
15796
15797
15798
15799
15800
15801
15802
15803
15804
15805
15806
15807
15808
15809
15810
15811
15812
15813
15814
15815
15816
15817
15818
15819
15820
15821
15822
15823
15824
15825
15826
15827
15828
15829
15830
15831
15832
15833
15834
15835
15836
15837
15838
15839
15840
15841
15842
15843
15844
15845
15846
15847
15848
15849
15850
15851
15852
15853
15854
15855
15856
15857
15858
15859
15860
15861
15862
15863
15864
15865
15866
15867
15868
15869
15870
15871
15872
15873
15874
15875
15876
15877
15878
15879
15880
15881
15882
15883
15884
15885
15886
15887
15888
15889
15890
15891
15892
15893
15894
15895
15896
15897
15898
15899
15900
15901
15902
15903
15904
15905
15906
15907
15908
15909
15910
15911
15912
15913
15914
15915
15916
15917
15918
15919
15920
15921
15922
15923
15924
15925
15926
15927
15928
15929
15930
15931
15932
15933
15934
15935
15936
15937
15938
15939
15940
15941
15942
15943
15944
15945
15946
15947
15948
15949
15950
15951
15952
15953
15954
15955
15956
15957
15958
15959
15960
15961
15962
15963
15964
15965
15966
15967
15968
15969
15970
15971
15972
15973
15974
15975
15976
15977
15978
15979
15980
15981
15982
15983
15984
15985
15986
15987
15988
15989
15990
15991
15992
15993
15994
15995
15996
15997
15998
15999
16000
16001
16002
16003
16004
16005
16006
16007
16008
16009
16010
16011
16012
16013
16014
16015
16016
16017
16018
16019
16020
16021
16022
16023
16024
16025
16026
16027
16028
16029
16030
16031
16032
16033
16034
16035
16036
16037
16038
16039
16040
16041
16042
16043
16044
16045
16046
16047
16048
16049
16050
16051
16052
16053
16054
16055
16056
16057
16058
16059
16060
16061
16062
16063
16064
16065
16066
16067
16068
16069
16070
16071
16072
16073
16074
16075
16076
16077
16078
16079
16080
16081
16082
16083
16084
16085
16086
16087
16088
16089
16090
16091
16092
16093
16094
16095
16096
16097
16098
16099
16100
16101
16102
16103
16104
16105
16106
16107
16108
16109
16110
16111
16112
16113
16114
16115
16116
16117
16118
16119
16120
16121
16122
16123
16124
16125
16126
16127
16128
16129
16130
16131
16132
16133
16134
16135
16136
16137
16138
16139
16140
16141
16142
16143
16144
16145
16146
16147
16148
16149
16150
16151
16152
16153
16154
16155
16156
16157
16158
16159
16160
16161
16162
16163
16164
16165
16166
16167
16168
16169
16170
16171
16172
16173
16174
16175
16176
16177
16178
16179
16180
16181
16182
16183
16184
16185
16186
16187
16188
16189
16190
16191
16192
16193
16194
16195
16196
16197
16198
16199
16200
16201
16202
16203
16204
16205
16206
16207
16208
16209
16210
16211
16212
16213
16214
16215
16216
16217
16218
16219
16220
16221
16222
16223
16224
16225
16226
16227
16228
16229
16230
16231
16232
16233
16234
16235
16236
16237
16238
16239
16240
16241
16242
16243
16244
16245
16246
16247
16248
16249
16250
16251
16252
16253
16254
16255
16256
16257
16258
16259
16260
16261
16262
16263
16264
16265
16266
16267
16268
16269
16270
16271
16272
16273
16274
16275
16276
16277
16278
16279
16280
16281
16282
16283
16284
16285
16286
16287
16288
16289
16290
16291
16292
16293
16294
16295
16296
16297
16298
16299
16300
16301
16302
16303
16304
16305
16306
16307
16308
16309
16310
16311
16312
16313
16314
16315
16316
16317
16318
16319
16320
16321
16322
16323
16324
16325
16326
16327
16328
16329
16330
16331
16332
16333
16334
16335
16336
16337
16338
16339
16340
16341
16342
16343
16344
16345
16346
16347
16348
16349
16350
16351
16352
16353
16354
16355
16356
16357
16358
16359
16360
16361
16362
16363
16364
16365
16366
16367
16368
16369
16370
16371
16372
16373
16374
16375
16376
16377
16378
16379
16380
16381
16382
16383
16384
16385
16386
16387
16388
16389
16390
16391
16392
16393
16394
16395
16396
16397
16398
16399
16400
16401
16402
16403
16404
16405
16406
16407
16408
16409
16410
16411
16412
16413
16414
16415
16416
16417
16418
16419
16420
16421
16422
16423
16424
16425
16426
16427
16428
16429
16430
16431
16432
16433
16434
16435
16436
16437
16438
16439
16440
16441
16442
16443
16444
16445
16446
16447
16448
16449
16450
16451
16452
16453
16454
16455
16456
16457
16458
16459
16460
16461
16462
16463
16464
16465
16466
16467
16468
16469
16470
16471
16472
16473
16474
16475
16476
16477
16478
16479
16480
16481
16482
16483
16484
16485
16486
16487
16488
16489
16490
16491
16492
16493
16494
16495
16496
16497
16498
16499
16500
16501
16502
16503
16504
16505
16506
16507
16508
16509
16510
16511
16512
16513
16514
16515
16516
16517
16518
16519
16520
16521
16522
16523
16524
16525
16526
16527
16528
16529
16530
16531
16532
16533
16534
16535
16536
16537
16538
16539
16540
16541
16542
16543
16544
16545
16546
16547
16548
16549
16550
16551
16552
16553
16554
16555
16556
16557
16558
16559
16560
16561
16562
16563
16564
16565
16566
16567
16568
16569
16570
16571
16572
16573
16574
16575
16576
16577
16578
16579
16580
16581
16582
16583
16584
16585
16586
16587
16588
16589
16590
16591
16592
16593
16594
16595
16596
16597
16598
16599
16600
16601
16602
16603
16604
16605
16606
16607
16608
16609
16610
16611
16612
16613
16614
16615
16616
16617
16618
16619
16620
16621
16622
16623
16624
16625
16626
16627
16628
16629
16630
16631
16632
16633
16634
16635
16636
16637
16638
16639
16640
16641
16642
16643
16644
16645
16646
16647
16648
16649
16650
16651
16652
16653
16654
16655
16656
16657
16658
16659
16660
16661
16662
16663
16664
16665
16666
16667
16668
16669
16670
16671
16672
16673
16674
16675
16676
16677
16678
16679
16680
16681
16682
16683
16684
16685
16686
16687
16688
16689
16690
16691
16692
16693
16694
16695
16696
16697
16698
16699
16700
16701
16702
16703
16704
16705
16706
16707
16708
16709
16710
16711
16712
16713
16714
16715
16716
16717
16718
16719
16720
16721
16722
16723
16724
16725
16726
16727
16728
16729
16730
16731
16732
16733
16734
16735
16736
16737
16738
16739
16740
16741
16742
16743
16744
16745
16746
16747
16748
16749
16750
16751
16752
16753
16754
16755
16756
16757
16758
16759
16760
16761
16762
16763
16764
16765
16766
16767
16768
16769
16770
16771
16772
16773
16774
16775
16776
16777
16778
16779
16780
16781
16782
16783
16784
16785
16786
16787
16788
16789
16790
16791
16792
16793
16794
16795
16796
16797
16798
16799
16800
16801
16802
16803
16804
16805
16806
16807
16808
16809
16810
16811
16812
16813
16814
16815
16816
16817
16818
16819
16820
16821
16822
16823
16824
16825
16826
16827
16828
16829
16830
16831
16832
16833
16834
16835
16836
16837
16838
16839
16840
16841
16842
16843
16844
16845
16846
16847
16848
16849
16850
16851
16852
16853
16854
16855
16856
16857
16858
16859
16860
16861
16862
16863
16864
16865
16866
16867
16868
16869
16870
16871
16872
16873
16874
16875
16876
16877
16878
16879
16880
16881
16882
16883
16884
16885
16886
16887
16888
16889
16890
16891
16892
16893
16894
16895
16896
16897
16898
16899
16900
16901
16902
16903
16904
16905
16906
16907
16908
16909
16910
16911
16912
16913
16914
16915
16916
16917
16918
16919
16920
16921
16922
16923
16924
16925
16926
16927
16928
16929
16930
16931
16932
16933
16934
16935
16936
16937
16938
16939
16940
16941
16942
16943
16944
16945
16946
16947
16948
16949
16950
16951
16952
16953
16954
16955
16956
16957
16958
16959
16960
16961
16962
16963
16964
16965
16966
16967
16968
16969
16970
16971
16972
16973
16974
16975
16976
16977
16978
16979
16980
16981
16982
16983
16984
16985
16986
16987
16988
16989
16990
16991
16992
16993
16994
16995
16996
16997
16998
16999
17000
17001
17002
17003
17004
17005
17006
17007
17008
17009
17010
17011
17012
17013
17014
17015
17016
17017
17018
17019
17020
17021
17022
17023
17024
17025
17026
17027
17028
17029
17030
17031
17032
17033
17034
17035
17036
17037
17038
17039
17040
17041
17042
17043
17044
17045
17046
17047
17048
17049
17050
17051
17052
17053
17054
17055
17056
17057
17058
17059
17060
17061
17062
17063
17064
17065
17066
17067
17068
17069
17070
17071
17072
17073
17074
17075
17076
17077
17078
17079
17080
17081
17082
17083
17084
17085
17086
17087
17088
17089
17090
17091
17092
17093
17094
17095
17096
17097
17098
17099
17100
17101
17102
17103
17104
17105
17106
17107
17108
17109
17110
17111
17112
17113
17114
17115
17116
17117
17118
17119
17120
17121
17122
17123
17124
17125
17126
17127
17128
17129
17130
17131
17132
17133
17134
17135
17136
17137
17138
17139
17140
17141
17142
17143
17144
17145
17146
17147
17148
17149
17150
17151
17152
17153
17154
17155
17156
17157
17158
17159
17160
17161
17162
17163
17164
17165
17166
17167
17168
17169
17170
17171
17172
17173
17174
17175
17176
17177
17178
17179
17180
17181
17182
17183
17184
17185
17186
17187
17188
17189
17190
17191
17192
17193
17194
17195
17196
17197
17198
17199
17200
17201
17202
17203
17204
17205
17206
17207
17208
17209
17210
17211
17212
17213
17214
17215
17216
17217
17218
17219
17220
17221
17222
17223
17224
17225
17226
17227
17228
17229
17230
17231
17232
17233
17234
17235
17236
17237
17238
17239
17240
17241
17242
17243
17244
17245
17246
17247
17248
17249
17250
17251
17252
17253
17254
17255
17256
17257
17258
17259
17260
17261
17262
17263
17264
17265
17266
17267
17268
17269
17270
17271
17272
17273
17274
17275
17276
17277
17278
17279
17280
17281
17282
17283
17284
17285
17286
17287
17288
17289
17290
17291
17292
17293
17294
17295
17296
17297
17298
17299
17300
17301
17302
17303
17304
17305
17306
17307
17308
17309
17310
17311
17312
17313
17314
17315
17316
17317
17318
17319
17320
17321
17322
17323
17324
17325
17326
17327
17328
17329
17330
17331
17332
17333
17334
17335
17336
17337
17338
17339
17340
17341
17342
17343
17344
17345
17346
17347
17348
17349
17350
17351
17352
17353
17354
17355
17356
17357
17358
17359
17360
17361
17362
17363
17364
17365
17366
17367
17368
17369
17370
17371
17372
17373
17374
17375
17376
17377
17378
17379
17380
17381
17382
17383
17384
17385
17386
17387
17388
17389
17390
17391
17392
17393
17394
17395
17396
17397
17398
17399
17400
17401
17402
17403
17404
17405
17406
17407
17408
17409
17410
17411
17412
17413
17414
17415
17416
17417
17418
17419
17420
17421
17422
17423
17424
17425
17426
17427
17428
17429
17430
17431
17432
17433
17434
17435
17436
17437
17438
17439
17440
17441
17442
17443
17444
17445
17446
17447
17448
17449
17450
17451
17452
17453
17454
17455
17456
17457
17458
17459
17460
17461
17462
17463
17464
17465
17466
17467
17468
17469
17470
17471
17472
17473
17474
17475
17476
17477
17478
17479
17480
17481
17482
17483
17484
17485
17486
17487
17488
17489
17490
17491
17492
17493
17494
17495
17496
17497
17498
17499
17500
17501
17502
17503
17504
17505
17506
17507
17508
17509
17510
17511
17512
17513
17514
17515
17516
17517
17518
17519
17520
17521
17522
17523
17524
17525
17526
17527
17528
17529
17530
17531
17532
17533
17534
17535
17536
17537
17538
17539
17540
17541
17542
17543
17544
17545
17546
17547
17548
17549
17550
17551
17552
17553
17554
17555
17556
17557
17558
17559
17560
17561
17562
17563
17564
17565
17566
17567
17568
17569
17570
17571
17572
17573
17574
17575
17576
17577
17578
17579
17580
17581
17582
17583
17584
17585
17586
17587
17588
17589
17590
17591
17592
17593
17594
17595
17596
17597
17598
17599
17600
17601
17602
17603
17604
17605
17606
17607
17608
17609
17610
17611
17612
17613
17614
17615
17616
17617
17618
17619
17620
17621
17622
17623
17624
17625
17626
17627
17628
17629
17630
17631
17632
17633
17634
17635
17636
17637
17638
17639
17640
17641
17642
17643
17644
17645
17646
17647
17648
17649
17650
17651
17652
17653
17654
17655
17656
17657
17658
17659
17660
17661
17662
17663
17664
17665
17666
17667
17668
17669
17670
17671
17672
17673
17674
17675
17676
17677
17678
17679
17680
17681
17682
17683
17684
17685
17686
17687
17688
17689
17690
17691
17692
17693
17694
17695
17696
17697
17698
17699
17700
17701
17702
17703
17704
17705
17706
17707
17708
17709
17710
17711
17712
17713
17714
17715
17716
17717
17718
17719
17720
17721
17722
17723
17724
17725
17726
17727
17728
17729
17730
17731
17732
17733
17734
17735
17736
17737
17738
17739
17740
17741
17742
17743
17744
17745
17746
17747
17748
17749
17750
17751
17752
17753
17754
17755
17756
17757
17758
17759
17760
17761
17762
17763
17764
17765
17766
17767
17768
17769
17770
17771
17772
17773
17774
17775
17776
17777
17778
17779
17780
17781
17782
17783
17784
17785
17786
17787
17788
17789
17790
17791
17792
17793
17794
17795
17796
17797
17798
17799
17800
17801
17802
17803
17804
17805
17806
17807
17808
17809
17810
17811
17812
17813
17814
17815
17816
17817
17818
17819
17820
17821
17822
17823
17824
17825
17826
17827
17828
17829
17830
17831
17832
17833
17834
17835
17836
17837
17838
17839
17840
17841
17842
17843
17844
17845
17846
17847
17848
17849
17850
17851
17852
17853
17854
17855
17856
17857
17858
17859
17860
17861
17862
17863
17864
17865
17866
17867
17868
17869
17870
17871
17872
17873
17874
17875
17876
17877
17878
17879
17880
17881
17882
17883
17884
17885
17886
17887
17888
17889
17890
17891
17892
17893
17894
17895
17896
17897
17898
17899
17900
17901
17902
17903
17904
17905
17906
17907
17908
17909
17910
17911
17912
17913
17914
17915
17916
17917
17918
17919
17920
17921
17922
17923
17924
17925
17926
17927
17928
17929
17930
17931
17932
17933
17934
17935
17936
17937
17938
17939
17940
17941
17942
17943
17944
17945
17946
17947
17948
17949
17950
17951
17952
17953
17954
17955
17956
17957
17958
17959
17960
17961
17962
17963
17964
17965
17966
17967
17968
17969
17970
17971
17972
17973
17974
17975
17976
17977
17978
17979
17980
17981
17982
17983
17984
17985
17986
17987
17988
17989
17990
17991
17992
17993
17994
17995
17996
17997
17998
17999
18000
18001
18002
18003
18004
18005
18006
18007
18008
18009
18010
18011
18012
18013
18014
18015
18016
18017
18018
18019
18020
18021
18022
18023
18024
18025
18026
18027
18028
18029
18030
18031
18032
18033
18034
18035
18036
18037
18038
18039
18040
18041
18042
18043
18044
18045
18046
18047
18048
18049
18050
18051
18052
18053
18054
18055
18056
18057
18058
18059
18060
18061
18062
18063
18064
18065
18066
18067
18068
18069
18070
18071
18072
18073
18074
18075
18076
18077
18078
18079
18080
18081
18082
18083
18084
18085
18086
18087
18088
18089
18090
18091
18092
18093
18094
18095
18096
18097
18098
18099
18100
18101
18102
18103
18104
18105
18106
18107
18108
18109
18110
18111
18112
18113
18114
18115
18116
18117
18118
18119
18120
18121
18122
18123
18124
18125
18126
18127
18128
18129
18130
18131
18132
18133
18134
18135
18136
18137
18138
18139
18140
18141
18142
18143
18144
18145
18146
18147
18148
18149
18150
18151
18152
18153
18154
18155
18156
18157
18158
18159
18160
18161
18162
18163
18164
18165
18166
18167
18168
18169
18170
18171
18172
18173
18174
18175
18176
18177
18178
18179
18180
18181
18182
18183
18184
18185
18186
18187
18188
18189
18190
18191
18192
18193
18194
18195
18196
18197
18198
18199
18200
18201
18202
18203
18204
18205
18206
18207
18208
18209
18210
18211
18212
18213
18214
18215
18216
18217
18218
18219
18220
18221
18222
18223
18224
18225
18226
18227
18228
18229
18230
18231
18232
18233
18234
18235
18236
18237
18238
18239
18240
18241
18242
18243
18244
18245
18246
18247
18248
18249
18250
18251
18252
18253
18254
18255
18256
18257
18258
18259
18260
18261
18262
18263
18264
18265
18266
18267
18268
18269
18270
18271
18272
18273
18274
18275
18276
18277
18278
18279
18280
18281
18282
18283
18284
18285
18286
18287
18288
18289
18290
18291
18292
18293
18294
18295
18296
18297
18298
18299
18300
18301
18302
18303
18304
18305
18306
18307
18308
18309
18310
18311
18312
18313
18314
18315
18316
18317
18318
18319
18320
18321
18322
18323
18324
18325
18326
18327
18328
18329
18330
18331
18332
18333
18334
18335
18336
18337
18338
18339
18340
18341
18342
18343
18344
18345
18346
18347
18348
18349
18350
18351
18352
18353
18354
18355
18356
18357
18358
18359
18360
18361
18362
18363
18364
18365
18366
18367
18368
18369
18370
18371
18372
18373
18374
18375
18376
18377
18378
18379
18380
18381
18382
18383
18384
18385
18386
18387
18388
18389
18390
18391
18392
18393
18394
18395
18396
18397
18398
18399
18400
18401
18402
18403
18404
18405
18406
18407
18408
18409
18410
18411
18412
18413
18414
18415
18416
18417
18418
18419
18420
18421
18422
18423
18424
18425
18426
18427
18428
18429
18430
18431
18432
18433
18434
18435
18436
18437
18438
18439
18440
18441
18442
18443
18444
18445
18446
18447
18448
18449
18450
18451
18452
18453
18454
18455
18456
18457
18458
18459
18460
18461
18462
18463
18464
18465
18466
18467
18468
18469
18470
18471
18472
18473
18474
18475
18476
18477
18478
18479
18480
18481
18482
18483
18484
18485
18486
18487
18488
18489
18490
18491
18492
18493
18494
18495
18496
18497
18498
18499
18500
18501
18502
18503
18504
18505
18506
18507
18508
18509
18510
18511
18512
18513
18514
18515
18516
18517
18518
18519
18520
18521
18522
18523
18524
18525
18526
18527
18528
18529
18530
18531
18532
18533
18534
18535
18536
18537
18538
18539
18540
18541
18542
18543
18544
18545
18546
18547
18548
18549
18550
18551
18552
18553
18554
18555
18556
18557
18558
18559
18560
18561
18562
18563
18564
18565
18566
18567
18568
18569
18570
18571
18572
18573
18574
18575
18576
18577
18578
18579
18580
18581
18582
18583
18584
18585
18586
18587
18588
18589
18590
18591
18592
18593
18594
18595
18596
18597
18598
18599
18600
18601
18602
18603
18604
18605
18606
18607
18608
18609
18610
18611
18612
18613
18614
18615
18616
18617
18618
18619
18620
18621
18622
18623
18624
18625
18626
18627
18628
18629
18630
18631
18632
18633
18634
18635
18636
18637
18638
18639
18640
18641
18642
18643
18644
18645
18646
18647
18648
18649
18650
18651
18652
18653
18654
18655
18656
18657
18658
18659
18660
18661
18662
18663
18664
18665
18666
18667
18668
18669
18670
18671
18672
18673
18674
18675
18676
18677
18678
18679
18680
18681
18682
18683
18684
18685
18686
18687
18688
18689
18690
18691
18692
18693
18694
18695
18696
18697
18698
18699
18700
18701
18702
18703
18704
18705
18706
18707
18708
18709
18710
18711
18712
18713
18714
18715
18716
18717
18718
18719
18720
18721
18722
18723
18724
18725
18726
18727
18728
18729
18730
18731
18732
18733
18734
18735
18736
18737
18738
18739
18740
18741
18742
18743
18744
18745
18746
18747
18748
18749
18750
18751
18752
18753
18754
18755
18756
18757
18758
18759
18760
18761
18762
18763
18764
18765
18766
18767
18768
18769
18770
18771
18772
18773
18774
18775
18776
18777
18778
18779
18780
18781
18782
18783
18784
18785
18786
18787
18788
18789
18790
18791
18792
18793
18794
18795
18796
18797
18798
18799
18800
18801
18802
18803
18804
18805
18806
18807
18808
18809
18810
18811
18812
18813
18814
18815
18816
18817
18818
18819
18820
18821
18822
18823
18824
18825
18826
18827
18828
18829
18830
18831
18832
18833
18834
18835
18836
18837
18838
18839
18840
18841
18842
18843
18844
18845
18846
18847
18848
18849
18850
18851
18852
18853
18854
18855
18856
18857
18858
18859
18860
18861
18862
18863
18864
18865
18866
18867
18868
18869
18870
18871
18872
18873
18874
18875
18876
18877
18878
18879
18880
18881
18882
18883
18884
18885
18886
18887
18888
18889
18890
18891
18892
18893
18894
18895
18896
18897
18898
18899
18900
18901
18902
18903
18904
18905
18906
18907
18908
18909
18910
18911
18912
18913
18914
18915
18916
18917
18918
18919
18920
18921
18922
18923
18924
18925
18926
18927
18928
18929
18930
18931
18932
18933
18934
18935
18936
18937
18938
18939
18940
18941
18942
18943
18944
18945
18946
18947
18948
18949
18950
18951
18952
18953
18954
18955
18956
18957
18958
18959
18960
18961
18962
18963
18964
18965
18966
18967
18968
18969
18970
18971
18972
18973
18974
18975
18976
18977
18978
18979
18980
18981
18982
18983
18984
18985
18986
18987
18988
18989
18990
18991
18992
18993
18994
18995
18996
18997
18998
18999
19000
19001
19002
19003
19004
19005
19006
19007
19008
19009
19010
19011
19012
19013
19014
19015
19016
19017
19018
19019
19020
19021
19022
19023
19024
19025
19026
19027
19028
19029
19030
19031
19032
19033
19034
19035
19036
19037
19038
19039
19040
19041
19042
19043
19044
19045
19046
19047
19048
19049
19050
19051
19052
19053
19054
19055
19056
19057
19058
19059
19060
19061
19062
19063
19064
19065
19066
19067
19068
19069
19070
19071
19072
19073
19074
19075
19076
19077
19078
19079
19080
19081
19082
19083
19084
19085
19086
19087
19088
19089
19090
19091
19092
19093
19094
19095
19096
19097
19098
19099
19100
19101
19102
19103
19104
19105
19106
19107
19108
19109
19110
19111
19112
19113
19114
19115
19116
19117
19118
19119
19120
19121
19122
19123
19124
19125
19126
19127
19128
19129
19130
19131
19132
19133
19134
19135
19136
19137
19138
19139
19140
19141
19142
19143
19144
19145
19146
19147
19148
19149
19150
19151
19152
19153
19154
19155
19156
19157
19158
19159
19160
19161
19162
19163
19164
19165
19166
19167
19168
19169
19170
19171
19172
19173
19174
19175
19176
19177
19178
19179
19180
19181
19182
19183
19184
19185
19186
19187
19188
19189
19190
19191
19192
19193
19194
19195
19196
19197
19198
19199
19200
19201
19202
19203
19204
19205
19206
19207
19208
19209
19210
19211
19212
19213
19214
19215
19216
19217
19218
19219
19220
19221
19222
19223
19224
19225
19226
19227
19228
19229
19230
19231
19232
19233
19234
19235
19236
19237
19238
19239
19240
19241
19242
19243
19244
19245
19246
19247
19248
19249
19250
19251
19252
19253
19254
19255
19256
19257
19258
19259
19260
19261
19262
19263
19264
19265
19266
19267
19268
19269
19270
19271
19272
19273
19274
19275
19276
19277
19278
19279
19280
19281
19282
19283
19284
19285
19286
19287
19288
19289
19290
19291
19292
19293
19294
19295
19296
19297
19298
19299
19300
19301
19302
19303
19304
19305
19306
19307
19308
19309
19310
19311
19312
19313
19314
19315
19316
19317
19318
19319
19320
19321
19322
19323
19324
19325
19326
19327
19328
19329
19330
19331
19332
19333
19334
19335
19336
19337
19338
19339
19340
19341
19342
19343
19344
19345
19346
19347
19348
19349
19350
19351
19352
19353
19354
19355
19356
19357
19358
19359
19360
19361
19362
19363
19364
19365
19366
19367
19368
19369
19370
19371
19372
19373
19374
19375
19376
19377
19378
19379
19380
19381
19382
19383
19384
19385
19386
19387
19388
19389
19390
19391
19392
19393
19394
19395
19396
19397
19398
19399
19400
19401
19402
19403
19404
19405
19406
19407
19408
19409
19410
19411
19412
19413
19414
19415
19416
19417
19418
19419
19420
19421
19422
19423
19424
19425
19426
19427
19428
19429
19430
19431
19432
19433
19434
19435
19436
19437
19438
19439
19440
19441
19442
19443
19444
19445
19446
19447
19448
19449
19450
19451
19452
19453
19454
19455
19456
19457
19458
19459
19460
19461
19462
19463
19464
19465
19466
19467
19468
19469
19470
19471
19472
19473
19474
19475
19476
19477
19478
19479
19480
19481
19482
19483
19484
19485
19486
19487
19488
19489
19490
19491
19492
19493
19494
19495
19496
19497
19498
19499
19500
19501
19502
19503
19504
19505
19506
19507
19508
19509
19510
19511
19512
19513
19514
19515
19516
19517
19518
19519
19520
19521
19522
19523
19524
19525
19526
19527
19528
19529
19530
19531
19532
19533
19534
19535
19536
19537
19538
19539
19540
19541
19542
19543
19544
19545
19546
19547
19548
19549
19550
19551
19552
19553
19554
19555
19556
19557
19558
19559
19560
19561
19562
19563
19564
19565
19566
19567
19568
19569
19570
19571
19572
19573
19574
19575
19576
19577
19578
19579
19580
19581
19582
19583
19584
19585
19586
19587
19588
19589
19590
19591
19592
19593
19594
19595
19596
19597
19598
19599
19600
19601
19602
19603
19604
19605
19606
19607
19608
19609
19610
19611
19612
19613
19614
19615
19616
19617
19618
19619
19620
19621
19622
19623
19624
19625
19626
19627
19628
19629
19630
19631
19632
19633
19634
19635
19636
19637
19638
19639
19640
19641
19642
19643
19644
19645
19646
19647
19648
19649
19650
19651
19652
19653
19654
19655
19656
19657
19658
19659
19660
19661
19662
19663
19664
19665
19666
19667
19668
19669
19670
19671
19672
19673
19674
19675
19676
19677
19678
19679
19680
19681
19682
19683
19684
19685
19686
19687
19688
19689
19690
19691
19692
19693
19694
19695
19696
19697
19698
19699
19700
19701
19702
19703
19704
19705
19706
19707
19708
19709
19710
19711
19712
19713
19714
19715
19716
19717
19718
19719
19720
19721
19722
19723
19724
19725
19726
19727
19728
19729
19730
19731
19732
19733
19734
19735
19736
19737
19738
19739
19740
19741
19742
19743
19744
19745
19746
19747
19748
19749
19750
19751
19752
19753
19754
19755
19756
19757
19758
19759
19760
19761
19762
19763
19764
19765
19766
19767
19768
19769
19770
19771
19772
19773
19774
19775
19776
19777
19778
19779
19780
19781
19782
19783
19784
19785
19786
19787
19788
19789
19790
19791
19792
19793
19794
19795
19796
19797
19798
19799
19800
19801
19802
19803
19804
19805
19806
19807
19808
19809
19810
19811
19812
19813
19814
19815
19816
19817
19818
19819
19820
19821
19822
19823
19824
19825
19826
19827
19828
19829
19830
19831
19832
19833
19834
19835
19836
19837
19838
19839
19840
19841
19842
19843
19844
19845
19846
19847
19848
19849
19850
19851
19852
19853
19854
19855
19856
19857
19858
19859
19860
19861
19862
19863
19864
19865
19866
19867
19868
19869
19870
19871
19872
19873
19874
19875
19876
19877
19878
19879
19880
19881
19882
19883
19884
19885
19886
19887
19888
19889
19890
19891
19892
19893
19894
19895
19896
19897
19898
19899
19900
19901
19902
19903
19904
19905
19906
19907
19908
19909
19910
19911
19912
19913
19914
19915
19916
19917
19918
19919
19920
19921
19922
19923
19924
19925
19926
19927
19928
19929
19930
19931
19932
19933
19934
19935
19936
19937
19938
19939
19940
19941
19942
19943
19944
19945
19946
19947
19948
19949
19950
19951
19952
19953
19954
19955
19956
19957
19958
19959
19960
19961
19962
19963
19964
19965
19966
19967
19968
19969
19970
19971
19972
19973
19974
19975
19976
19977
19978
19979
19980
19981
19982
19983
19984
19985
19986
19987
19988
19989
19990
19991
19992
19993
19994
19995
19996
19997
19998
19999
20000
20001
20002
20003
20004
20005
20006
20007
20008
20009
20010
20011
20012
20013
20014
20015
20016
20017
20018
20019
20020
20021
20022
20023
20024
20025
20026
20027
20028
20029
20030
20031
20032
20033
20034
20035
20036
20037
20038
20039
20040
20041
20042
20043
20044
20045
20046
20047
20048
20049
20050
20051
20052
20053
20054
20055
20056
20057
20058
20059
20060
20061
20062
20063
20064
20065
20066
20067
20068
20069
20070
20071
20072
20073
20074
20075
20076
20077
20078
20079
20080
20081
20082
20083
20084
20085
20086
20087
20088
20089
20090
20091
20092
20093
20094
20095
20096
20097
20098
20099
20100
20101
20102
20103
20104
20105
20106
20107
20108
20109
20110
20111
20112
20113
20114
20115
20116
20117
20118
20119
20120
20121
20122
20123
20124
20125
20126
20127
20128
20129
20130
20131
20132
20133
20134
20135
20136
20137
20138
20139
20140
20141
20142
20143
20144
20145
20146
20147
20148
20149
20150
20151
20152
20153
20154
20155
20156
20157
20158
20159
20160
20161
20162
20163
20164
20165
20166
20167
20168
20169
20170
20171
20172
20173
20174
20175
20176
20177
20178
20179
20180
20181
20182
20183
20184
20185
20186
20187
20188
20189
20190
20191
20192
20193
20194
20195
20196
20197
20198
20199
20200
20201
20202
20203
20204
20205
20206
20207
20208
20209
20210
20211
20212
20213
20214
20215
20216
20217
20218
20219
20220
20221
20222
20223
20224
20225
20226
20227
20228
20229
20230
20231
20232
20233
20234
20235
20236
20237
20238
20239
20240
20241
20242
20243
20244
20245
20246
20247
20248
20249
20250
20251
20252
20253
20254
20255
20256
20257
20258
20259
20260
20261
20262
20263
20264
20265
20266
20267
20268
20269
20270
20271
20272
20273
20274
20275
20276
20277
20278
20279
20280
20281
20282
20283
20284
20285
20286
20287
20288
20289
20290
20291
20292
20293
20294
20295
20296
20297
20298
20299
20300
20301
20302
20303
20304
20305
20306
20307
20308
20309
20310
20311
20312
20313
20314
20315
20316
20317
20318
20319
20320
20321
20322
20323
20324
20325
20326
20327
20328
20329
20330
20331
20332
20333
20334
20335
20336
20337
20338
20339
20340
20341
20342
20343
20344
20345
20346
20347
20348
20349
20350
20351
20352
20353
20354
20355
20356
20357
20358
20359
20360
20361
20362
20363
20364
20365
20366
20367
20368
20369
20370
20371
20372
20373
20374
20375
20376
20377
20378
20379
20380
20381
20382
20383
20384
20385
20386
20387
20388
20389
20390
20391
20392
20393
20394
20395
20396
20397
20398
20399
20400
20401
20402
20403
20404
20405
20406
20407
20408
20409
20410
20411
20412
20413
20414
20415
20416
20417
20418
20419
20420
20421
20422
20423
20424
20425
20426
20427
20428
20429
20430
20431
20432
20433
20434
20435
20436
20437
20438
20439
20440
20441
20442
20443
20444
20445
20446
20447
20448
20449
20450
20451
20452
20453
20454
20455
20456
20457
20458
20459
20460
20461
20462
20463
20464
20465
20466
20467
20468
20469
20470
20471
20472
20473
20474
20475
20476
20477
20478
20479
20480
20481
20482
20483
20484
20485
20486
20487
20488
20489
20490
20491
20492
20493
20494
20495
20496
20497
20498
20499
20500
20501
20502
20503
20504
20505
20506
20507
20508
20509
20510
20511
20512
20513
20514
20515
20516
20517
20518
20519
20520
20521
20522
20523
20524
20525
20526
20527
20528
20529
20530
20531
20532
20533
20534
20535
20536
20537
20538
20539
20540
20541
20542
20543
20544
20545
20546
20547
20548
20549
20550
20551
20552
20553
20554
20555
20556
20557
20558
20559
20560
20561
20562
20563
20564
20565
20566
20567
20568
20569
20570
20571
20572
20573
20574
20575
20576
20577
20578
20579
20580
20581
20582
20583
20584
20585
20586
20587
20588
20589
20590
20591
20592
20593
20594
20595
20596
20597
20598
20599
20600
20601
20602
20603
20604
20605
20606
20607
20608
20609
20610
20611
20612
20613
20614
20615
20616
20617
20618
20619
20620
20621
20622
20623
20624
20625
20626
20627
20628
20629
20630
20631
20632
20633
20634
20635
20636
20637
20638
20639
20640
20641
20642
20643
20644
20645
20646
20647
20648
20649
20650
20651
20652
20653
20654
20655
20656
20657
20658
20659
20660
20661
20662
20663
20664
20665
20666
20667
20668
20669
20670
20671
20672
20673
20674
20675
20676
20677
20678
20679
20680
20681
20682
20683
20684
20685
20686
20687
20688
20689
20690
20691
20692
20693
20694
20695
20696
20697
20698
20699
20700
20701
20702
20703
20704
20705
20706
20707
20708
20709
20710
20711
20712
20713
20714
20715
20716
20717
20718
20719
20720
20721
20722
20723
20724
20725
20726
20727
20728
20729
20730
20731
20732
20733
20734
20735
20736
20737
20738
20739
20740
20741
20742
20743
20744
20745
20746
20747
20748
20749
20750
20751
20752
20753
20754
20755
20756
20757
20758
20759
20760
20761
20762
20763
20764
20765
20766
20767
20768
20769
20770
20771
20772
20773
20774
20775
20776
20777
20778
20779
20780
20781
20782
20783
20784
20785
20786
20787
20788
20789
20790
20791
20792
20793
20794
20795
20796
20797
20798
20799
20800
20801
20802
20803
20804
20805
20806
20807
20808
20809
20810
20811
20812
20813
20814
20815
20816
20817
20818
20819
20820
20821
20822
20823
20824
20825
20826
20827
20828
20829
20830
20831
20832
20833
20834
20835
20836
20837
20838
20839
20840
20841
20842
20843
20844
20845
20846
20847
20848
20849
20850
20851
20852
20853
20854
20855
20856
20857
20858
20859
20860
20861
20862
20863
20864
20865
20866
20867
20868
20869
20870
20871
20872
20873
20874
20875
20876
20877
20878
20879
20880
20881
20882
20883
20884
20885
20886
20887
20888
20889
20890
20891
20892
20893
20894
20895
20896
20897
20898
20899
20900
20901
20902
20903
20904
20905
20906
20907
20908
20909
20910
20911
20912
20913
20914
20915
20916
20917
20918
20919
20920
20921
20922
20923
20924
20925
20926
20927
20928
20929
20930
20931
20932
20933
20934
20935
20936
20937
20938
20939
20940
20941
20942
20943
20944
20945
20946
20947
20948
20949
20950
20951
20952
20953
20954
20955
20956
20957
20958
20959
20960
20961
20962
20963
20964
20965
20966
20967
20968
20969
20970
20971
20972
20973
20974
20975
20976
20977
20978
20979
20980
20981
20982
20983
20984
20985
20986
20987
20988
20989
20990
20991
20992
20993
20994
20995
20996
20997
20998
20999
21000
21001
21002
21003
21004
21005
21006
21007
21008
21009
21010
21011
21012
21013
21014
21015
21016
21017
21018
21019
21020
21021
21022
21023
21024
21025
21026
21027
21028
21029
21030
21031
21032
21033
21034
21035
21036
21037
21038
21039
21040
21041
21042
21043
21044
21045
21046
21047
21048
21049
21050
21051
21052
21053
21054
21055
21056
21057
21058
21059
21060
21061
21062
21063
21064
21065
21066
21067
21068
21069
21070
21071
21072
21073
21074
21075
21076
21077
21078
21079
21080
21081
21082
21083
21084
21085
21086
21087
21088
21089
21090
21091
21092
21093
21094
21095
21096
21097
21098
21099
21100
21101
21102
21103
21104
21105
21106
21107
21108
21109
21110
21111
21112
21113
21114
21115
21116
21117
21118
21119
21120
21121
21122
21123
21124
21125
21126
21127
21128
21129
21130
21131
21132
21133
21134
21135
21136
21137
21138
21139
21140
21141
21142
21143
21144
21145
21146
21147
21148
21149
21150
21151
21152
21153
21154
21155
21156
21157
21158
21159
21160
21161
21162
21163
21164
21165
21166
21167
21168
21169
21170
21171
21172
21173
21174
21175
21176
21177
21178
21179
21180
21181
21182
21183
21184
21185
21186
21187
21188
21189
21190
21191
21192
21193
21194
21195
21196
21197
21198
21199
21200
21201
21202
21203
21204
21205
21206
21207
21208
21209
21210
21211
21212
21213
21214
21215
21216
21217
21218
21219
21220
21221
21222
21223
21224
21225
21226
21227
21228
21229
21230
21231
21232
21233
21234
21235
21236
21237
21238
21239
21240
21241
21242
21243
21244
21245
21246
21247
21248
21249
21250
21251
21252
21253
21254
21255
21256
21257
21258
21259
21260
21261
21262
21263
21264
21265
21266
21267
21268
21269
21270
21271
21272
21273
21274
21275
21276
21277
21278
21279
21280
21281
21282
21283
21284
21285
21286
21287
21288
21289
21290
21291
21292
21293
21294
21295
21296
21297
21298
21299
21300
21301
21302
21303
21304
21305
21306
21307
21308
21309
21310
21311
21312
21313
21314
21315
21316
21317
21318
21319
21320
21321
21322
21323
21324
21325
21326
21327
21328
21329
21330
21331
21332
21333
21334
21335
21336
21337
21338
21339
21340
21341
21342
21343
21344
21345
21346
21347
21348
21349
21350
21351
21352
21353
21354
21355
21356
21357
21358
21359
21360
21361
21362
21363
21364
21365
21366
21367
21368
21369
21370
21371
21372
21373
21374
21375
21376
21377
21378
21379
21380
21381
21382
21383
21384
21385
21386
21387
21388
21389
21390
21391
21392
21393
21394
21395
21396
21397
21398
21399
21400
21401
21402
21403
21404
21405
21406
21407
21408
21409
21410
21411
21412
21413
21414
21415
21416
21417
21418
21419
21420
21421
21422
21423
21424
21425
21426
21427
21428
21429
21430
21431
21432
21433
21434
21435
21436
21437
21438
21439
21440
21441
21442
21443
21444
21445
21446
21447
21448
21449
21450
21451
21452
21453
21454
21455
21456
21457
21458
21459
21460
21461
21462
21463
21464
21465
21466
21467
21468
21469
21470
21471
21472
21473
21474
21475
21476
21477
21478
21479
21480
21481
21482
21483
21484
21485
21486
21487
21488
21489
21490
21491
21492
21493
21494
21495
21496
21497
21498
21499
21500
21501
21502
21503
21504
21505
21506
21507
21508
21509
21510
21511
21512
21513
21514
21515
21516
21517
21518
21519
21520
21521
21522
21523
21524
21525
21526
21527
21528
21529
21530
21531
21532
21533
21534
21535
21536
21537
21538
21539
21540
21541
21542
21543
21544
21545
21546
21547
21548
21549
21550
21551
21552
21553
21554
21555
21556
21557
21558
21559
21560
21561
21562
21563
21564
21565
21566
21567
21568
21569
21570
21571
21572
21573
21574
21575
21576
21577
21578
21579
21580
21581
21582
21583
21584
21585
21586
21587
21588
21589
21590
21591
21592
21593
21594
21595
21596
21597
21598
21599
21600
21601
21602
21603
21604
21605
21606
21607
21608
21609
21610
21611
21612
21613
21614
21615
21616
21617
21618
21619
21620
21621
21622
21623
21624
21625
21626
21627
21628
21629
21630
21631
21632
21633
21634
21635
21636
21637
21638
21639
21640
21641
21642
21643
21644
21645
21646
21647
21648
21649
21650
21651
21652
21653
21654
21655
21656
21657
21658
21659
21660
21661
21662
21663
21664
21665
21666
21667
21668
21669
21670
21671
21672
21673
21674
21675
21676
21677
21678
21679
21680
21681
21682
21683
21684
21685
21686
21687
21688
21689
21690
21691
21692
21693
21694
21695
21696
21697
21698
21699
21700
21701
21702
21703
21704
21705
21706
21707
21708
21709
21710
21711
21712
21713
21714
21715
21716
21717
21718
21719
21720
21721
21722
21723
21724
21725
21726
21727
21728
21729
21730
21731
21732
21733
21734
21735
21736
21737
21738
21739
21740
21741
21742
21743
21744
21745
21746
21747
21748
21749
21750
21751
21752
21753
21754
21755
21756
21757
21758
21759
21760
21761
21762
21763
21764
21765
21766
21767
21768
21769
21770
21771
21772
21773
21774
21775
21776
21777
21778
21779
21780
21781
21782
21783
21784
21785
21786
21787
21788
21789
21790
21791
21792
21793
21794
21795
21796
21797
21798
21799
21800
21801
21802
21803
21804
21805
21806
21807
21808
21809
21810
21811
21812
21813
21814
21815
21816
21817
21818
21819
21820
21821
21822
21823
21824
21825
21826
21827
21828
21829
21830
21831
21832
21833
21834
21835
21836
21837
21838
21839
21840
21841
21842
21843
21844
21845
21846
21847
21848
21849
21850
21851
21852
21853
21854
21855
21856
21857
21858
21859
21860
21861
21862
21863
21864
21865
21866
21867
21868
21869
21870
21871
21872
21873
21874
21875
21876
21877
21878
21879
21880
21881
21882
21883
21884
21885
21886
21887
21888
21889
21890
21891
21892
21893
21894
21895
21896
21897
21898
21899
21900
21901
21902
21903
21904
21905
21906
21907
21908
21909
21910
21911
21912
21913
21914
21915
21916
21917
21918
21919
21920
21921
21922
21923
21924
21925
21926
21927
21928
21929
21930
21931
21932
21933
21934
21935
21936
21937
21938
21939
21940
21941
21942
21943
21944
21945
21946
21947
21948
21949
21950
21951
21952
21953
21954
21955
21956
21957
21958
21959
21960
21961
21962
21963
21964
21965
21966
21967
21968
21969
21970
21971
21972
21973
21974
21975
21976
21977
21978
21979
21980
21981
21982
21983
21984
21985
21986
21987
21988
21989
21990
21991
21992
21993
21994
21995
21996
21997
21998
21999
22000
22001
22002
22003
22004
22005
22006
22007
22008
22009
22010
22011
22012
22013
22014
22015
22016
22017
22018
22019
22020
22021
22022
22023
22024
22025
22026
22027
22028
22029
22030
22031
22032
22033
22034
22035
22036
22037
22038
22039
22040
22041
22042
22043
22044
22045
22046
22047
22048
22049
22050
22051
22052
22053
22054
22055
22056
22057
22058
22059
22060
22061
22062
22063
22064
22065
22066
22067
22068
22069
22070
22071
22072
22073
22074
22075
22076
22077
22078
22079
22080
22081
22082
22083
22084
22085
22086
22087
22088
22089
22090
22091
22092
22093
22094
22095
22096
22097
22098
22099
22100
22101
22102
22103
22104
22105
22106
22107
22108
22109
22110
22111
22112
22113
22114
22115
22116
22117
22118
22119
22120
22121
22122
22123
22124
22125
22126
22127
22128
22129
22130
22131
22132
22133
22134
22135
22136
22137
22138
22139
22140
22141
22142
22143
22144
22145
22146
22147
22148
22149
22150
22151
22152
22153
22154
22155
22156
22157
22158
22159
22160
22161
22162
22163
22164
22165
22166
22167
22168
22169
22170
22171
22172
22173
22174
22175
22176
22177
22178
22179
22180
22181
22182
22183
22184
22185
22186
22187
22188
22189
22190
22191
22192
22193
22194
22195
22196
22197
22198
22199
22200
22201
22202
22203
22204
22205
22206
22207
22208
22209
22210
22211
22212
22213
22214
22215
22216
22217
22218
22219
22220
22221
22222
22223
22224
22225
22226
22227
22228
22229
22230
22231
22232
22233
22234
22235
22236
22237
22238
22239
22240
22241
22242
22243
22244
22245
22246
22247
22248
22249
22250
22251
22252
22253
22254
22255
22256
22257
22258
22259
22260
22261
22262
22263
22264
22265
22266
22267
22268
22269
22270
22271
22272
22273
22274
22275
22276
22277
22278
22279
22280
22281
22282
22283
22284
22285
22286
22287
22288
22289
22290
22291
22292
22293
22294
22295
22296
22297
22298
22299
22300
22301
22302
22303
22304
22305
22306
22307
22308
22309
22310
22311
22312
22313
22314
22315
22316
22317
22318
22319
22320
22321
22322
22323
22324
22325
22326
22327
22328
22329
22330
22331
22332
22333
22334
22335
22336
22337
22338
22339
22340
22341
22342
22343
22344
22345
22346
22347
22348
22349
22350
22351
22352
22353
22354
22355
22356
22357
22358
22359
22360
22361
22362
22363
22364
22365
22366
22367
22368
22369
22370
22371
22372
22373
22374
22375
22376
22377
22378
22379
22380
22381
22382
22383
22384
22385
22386
22387
22388
22389
22390
22391
22392
22393
22394
22395
22396
22397
22398
22399
22400
22401
22402
22403
22404
22405
22406
22407
22408
22409
22410
22411
22412
22413
22414
22415
22416
22417
22418
22419
22420
22421
22422
22423
22424
22425
22426
22427
22428
22429
22430
22431
22432
22433
22434
22435
22436
22437
22438
22439
22440
22441
22442
22443
22444
22445
22446
22447
22448
22449
22450
22451
22452
22453
22454
22455
22456
22457
22458
22459
22460
22461
22462
22463
22464
22465
22466
22467
22468
22469
22470
22471
22472
22473
22474
22475
22476
22477
22478
22479
22480
22481
22482
22483
22484
22485
22486
22487
22488
22489
22490
22491
22492
22493
22494
22495
22496
22497
22498
22499
22500
22501
22502
22503
22504
22505
22506
22507
22508
22509
22510
22511
22512
22513
22514
22515
22516
22517
22518
22519
22520
22521
22522
22523
22524
22525
22526
22527
22528
22529
22530
22531
22532
22533
22534
22535
22536
22537
22538
22539
22540
22541
22542
22543
22544
22545
22546
22547
22548
22549
22550
22551
22552
22553
22554
22555
22556
22557
22558
22559
22560
22561
22562
22563
22564
22565
22566
22567
22568
22569
22570
22571
22572
22573
22574
22575
22576
22577
22578
22579
22580
22581
22582
22583
22584
22585
22586
22587
22588
22589
22590
22591
22592
22593
22594
22595
22596
22597
22598
22599
22600
22601
22602
22603
22604
22605
22606
22607
22608
22609
22610
22611
22612
22613
22614
22615
22616
22617
22618
22619
22620
22621
22622
22623
22624
22625
22626
22627
22628
22629
22630
22631
22632
22633
22634
22635
22636
22637
22638
22639
22640
22641
22642
22643
22644
22645
22646
22647
22648
22649
22650
22651
22652
22653
22654
22655
22656
22657
22658
22659
22660
22661
22662
22663
22664
22665
22666
22667
22668
22669
22670
22671
22672
22673
22674
22675
22676
22677
22678
22679
22680
22681
22682
22683
22684
22685
22686
22687
22688
22689
22690
22691
22692
22693
22694
22695
22696
22697
22698
22699
22700
22701
22702
22703
22704
22705
22706
22707
22708
22709
22710
22711
22712
22713
22714
22715
22716
22717
22718
22719
22720
22721
22722
22723
22724
22725
22726
22727
22728
22729
22730
22731
22732
22733
22734
22735
22736
22737
22738
22739
22740
22741
22742
22743
22744
22745
22746
22747
22748
22749
22750
22751
22752
22753
22754
22755
22756
22757
22758
22759
22760
22761
22762
22763
22764
22765
22766
22767
22768
22769
22770
22771
22772
22773
22774
22775
22776
22777
22778
22779
22780
22781
22782
22783
22784
22785
22786
22787
22788
22789
22790
22791
22792
22793
22794
22795
22796
22797
22798
22799
22800
22801
22802
22803
22804
22805
22806
22807
22808
22809
22810
22811
22812
22813
22814
22815
22816
22817
22818
22819
22820
22821
22822
22823
22824
22825
22826
22827
22828
22829
22830
22831
22832
22833
22834
22835
22836
22837
22838
22839
22840
22841
22842
22843
22844
22845
22846
22847
22848
22849
22850
22851
22852
22853
22854
22855
22856
22857
22858
22859
22860
22861
22862
22863
22864
22865
22866
22867
22868
22869
22870
22871
22872
22873
22874
22875
22876
22877
22878
22879
22880
22881
22882
22883
22884
22885
22886
22887
22888
22889
22890
22891
22892
22893
22894
22895
22896
22897
22898
22899
22900
22901
22902
22903
22904
22905
22906
22907
22908
22909
22910
22911
22912
22913
22914
22915
22916
22917
22918
22919
22920
22921
22922
22923
22924
22925
22926
22927
22928
22929
22930
22931
22932
22933
22934
22935
22936
22937
22938
22939
22940
22941
22942
22943
22944
22945
22946
22947
22948
22949
22950
22951
22952
22953
22954
22955
22956
22957
22958
22959
22960
22961
22962
22963
22964
22965
22966
22967
22968
22969
22970
22971
22972
22973
22974
22975
22976
22977
22978
22979
22980
22981
22982
22983
22984
22985
22986
22987
22988
22989
22990
22991
22992
22993
22994
22995
22996
22997
22998
22999
23000
23001
23002
23003
23004
23005
23006
23007
23008
23009
23010
23011
23012
23013
23014
23015
23016
23017
23018
23019
23020
23021
23022
23023
23024
23025
23026
23027
23028
23029
23030
23031
23032
23033
23034
23035
23036
23037
23038
23039
23040
23041
23042
23043
23044
23045
23046
23047
23048
23049
23050
23051
23052
23053
23054
23055
23056
23057
23058
23059
23060
23061
23062
23063
23064
23065
23066
23067
23068
23069
23070
23071
23072
23073
23074
23075
23076
23077
23078
23079
23080
23081
23082
23083
23084
23085
23086
23087
23088
23089
23090
23091
23092
23093
23094
23095
23096
23097
23098
23099
23100
23101
23102
23103
23104
23105
23106
23107
23108
23109
23110
23111
23112
23113
23114
23115
23116
23117
23118
23119
23120
23121
23122
23123
23124
23125
23126
23127
23128
23129
23130
23131
23132
23133
23134
23135
23136
23137
23138
23139
23140
23141
23142
23143
23144
23145
23146
23147
23148
23149
23150
23151
23152
23153
23154
23155
23156
23157
23158
23159
23160
23161
23162
23163
23164
23165
23166
23167
23168
23169
23170
23171
23172
23173
23174
23175
23176
23177
23178
23179
23180
23181
23182
23183
23184
23185
23186
23187
23188
23189
23190
23191
23192
23193
23194
23195
23196
23197
23198
23199
23200
23201
23202
23203
23204
23205
23206
23207
23208
23209
23210
23211
23212
23213
23214
23215
23216
23217
23218
23219
23220
23221
23222
23223
23224
23225
23226
23227
23228
23229
23230
23231
23232
23233
23234
23235
23236
23237
23238
23239
23240
23241
23242
23243
23244
23245
23246
23247
23248
23249
23250
23251
23252
23253
23254
23255
23256
23257
23258
23259
23260
23261
23262
23263
23264
23265
23266
23267
23268
23269
23270
23271
23272
23273
23274
23275
23276
23277
23278
23279
23280
23281
23282
23283
23284
23285
23286
23287
23288
23289
23290
23291
23292
23293
23294
23295
23296
23297
23298
23299
23300
23301
23302
23303
23304
23305
23306
23307
23308
23309
23310
23311
23312
23313
23314
23315
23316
23317
23318
23319
23320
23321
23322
23323
23324
23325
23326
23327
23328
23329
23330
23331
23332
23333
23334
23335
23336
23337
23338
23339
23340
23341
23342
23343
23344
23345
23346
23347
23348
23349
23350
23351
23352
23353
23354
23355
23356
23357
23358
23359
23360
23361
23362
23363
23364
23365
23366
23367
23368
23369
23370
23371
23372
23373
23374
23375
23376
23377
23378
23379
23380
23381
23382
23383
23384
23385
23386
23387
23388
23389
23390
23391
23392
23393
23394
23395
23396
23397
23398
23399
23400
23401
23402
23403
23404
23405
23406
23407
23408
23409
23410
23411
23412
23413
23414
23415
23416
23417
23418
23419
23420
23421
23422
23423
23424
23425
23426
23427
23428
23429
23430
23431
23432
23433
23434
23435
23436
23437
23438
23439
23440
23441
23442
23443
23444
23445
23446
23447
23448
23449
23450
23451
23452
23453
23454
23455
23456
23457
23458
23459
23460
23461
23462
23463
23464
23465
23466
23467
23468
23469
23470
23471
23472
23473
23474
23475
23476
23477
23478
23479
23480
23481
23482
23483
23484
23485
23486
23487
23488
23489
23490
23491
23492
23493
23494
23495
23496
23497
23498
23499
23500
23501
23502
23503
23504
23505
23506
23507
23508
23509
23510
23511
23512
23513
23514
23515
23516
23517
23518
23519
23520
23521
23522
23523
23524
23525
23526
23527
23528
23529
23530
23531
23532
23533
23534
23535
23536
23537
23538
23539
23540
23541
23542
23543
23544
23545
23546
23547
23548
23549
23550
23551
23552
23553
23554
23555
23556
23557
23558
23559
23560
23561
23562
23563
23564
23565
23566
23567
23568
23569
23570
23571
23572
23573
23574
23575
23576
23577
23578
23579
23580
23581
23582
23583
23584
23585
23586
23587
23588
23589
23590
23591
23592
23593
23594
23595
23596
23597
23598
23599
23600
23601
23602
23603
23604
23605
23606
23607
23608
23609
23610
23611
23612
23613
23614
23615
23616
23617
23618
23619
23620
23621
23622
23623
23624
23625
23626
23627
23628
23629
23630
23631
23632
23633
23634
23635
23636
23637
23638
23639
23640
23641
23642
23643
23644
23645
23646
23647
23648
23649
23650
23651
23652
23653
23654
23655
23656
23657
23658
23659
23660
23661
23662
23663
23664
23665
23666
23667
23668
23669
23670
23671
23672
23673
23674
23675
23676
23677
23678
23679
23680
23681
23682
23683
23684
23685
23686
23687
23688
23689
23690
23691
23692
23693
23694
23695
23696
23697
23698
23699
23700
23701
23702
23703
23704
23705
23706
23707
23708
23709
23710
23711
23712
23713
23714
23715
23716
23717
23718
23719
23720
23721
23722
23723
23724
23725
23726
23727
23728
23729
23730
23731
23732
23733
23734
23735
23736
23737
23738
23739
23740
23741
23742
23743
23744
23745
23746
23747
23748
23749
23750
23751
23752
23753
23754
23755
23756
23757
23758
23759
23760
23761
23762
23763
23764
23765
23766
23767
23768
23769
23770
23771
23772
23773
23774
23775
23776
23777
23778
23779
23780
23781
23782
23783
23784
23785
23786
23787
23788
23789
23790
23791
23792
23793
23794
23795
23796
23797
23798
23799
23800
23801
23802
23803
23804
23805
23806
23807
23808
23809
23810
23811
23812
23813
23814
23815
23816
23817
23818
23819
23820
23821
23822
23823
23824
23825
23826
23827
23828
23829
23830
23831
23832
23833
23834
23835
23836
23837
23838
23839
23840
23841
23842
23843
23844
23845
23846
23847
23848
23849
23850
23851
23852
23853
23854
23855
23856
23857
23858
23859
23860
23861
23862
23863
23864
23865
23866
23867
23868
23869
23870
23871
23872
23873
23874
23875
23876
23877
23878
23879
23880
23881
23882
23883
23884
23885
23886
23887
23888
23889
23890
23891
23892
23893
23894
23895
23896
23897
23898
23899
23900
23901
23902
23903
23904
23905
23906
23907
23908
23909
23910
23911
23912
23913
23914
23915
23916
23917
23918
23919
23920
23921
23922
23923
23924
23925
23926
23927
23928
23929
23930
23931
23932
23933
23934
23935
23936
23937
23938
23939
23940
23941
23942
23943
23944
23945
23946
23947
23948
23949
23950
23951
23952
23953
23954
23955
23956
23957
23958
23959
23960
23961
23962
23963
23964
23965
23966
23967
23968
23969
23970
23971
23972
23973
23974
23975
23976
23977
23978
23979
23980
23981
23982
23983
23984
23985
23986
23987
23988
23989
23990
23991
23992
23993
23994
23995
23996
23997
23998
23999
24000
24001
24002
24003
24004
24005
24006
24007
24008
24009
24010
24011
24012
24013
24014
24015
24016
24017
24018
24019
24020
24021
24022
24023
24024
24025
24026
24027
24028
24029
24030
24031
24032
24033
24034
24035
24036
24037
24038
24039
24040
24041
24042
24043
24044
24045
24046
24047
24048
24049
24050
24051
24052
24053
24054
24055
24056
24057
24058
24059
24060
24061
24062
24063
24064
24065
24066
24067
24068
24069
24070
24071
24072
24073
24074
24075
24076
24077
24078
24079
24080
24081
24082
24083
24084
24085
24086
24087
24088
24089
24090
24091
24092
24093
24094
24095
24096
24097
24098
24099
24100
24101
24102
24103
24104
24105
24106
24107
24108
24109
24110
24111
24112
24113
24114
24115
24116
24117
24118
24119
24120
24121
24122
24123
24124
24125
24126
24127
24128
24129
24130
24131
24132
24133
24134
24135
24136
24137
24138
24139
24140
24141
24142
24143
24144
24145
24146
24147
24148
24149
24150
24151
24152
24153
24154
24155
24156
24157
24158
24159
24160
24161
24162
24163
24164
24165
24166
24167
24168
24169
24170
24171
24172
24173
24174
24175
24176
24177
24178
24179
24180
24181
24182
24183
24184
24185
24186
24187
24188
24189
24190
24191
24192
24193
24194
24195
24196
24197
24198
24199
24200
24201
24202
24203
24204
24205
24206
24207
24208
24209
24210
24211
24212
24213
24214
24215
24216
24217
24218
24219
24220
24221
24222
24223
24224
24225
24226
24227
24228
24229
24230
24231
24232
24233
24234
24235
24236
24237
24238
24239
24240
24241
24242
24243
24244
24245
24246
24247
24248
24249
24250
24251
24252
24253
24254
24255
24256
24257
24258
24259
24260
24261
24262
24263
24264
24265
24266
24267
24268
24269
24270
24271
24272
24273
24274
24275
24276
24277
24278
24279
24280
24281
24282
24283
24284
24285
24286
24287
24288
24289
24290
24291
24292
24293
24294
24295
24296
24297
24298
24299
24300
24301
24302
24303
24304
24305
24306
24307
24308
24309
24310
24311
24312
24313
24314
24315
24316
24317
24318
24319
24320
24321
24322
24323
24324
24325
24326
24327
24328
24329
24330
24331
24332
24333
24334
24335
24336
24337
24338
24339
24340
24341
24342
24343
24344
24345
24346
24347
24348
24349
24350
24351
24352
24353
24354
24355
24356
24357
24358
24359
24360
24361
24362
24363
24364
24365
24366
24367
24368
24369
24370
24371
24372
24373
24374
24375
24376
24377
24378
24379
24380
24381
24382
24383
24384
24385
24386
24387
24388
24389
24390
24391
24392
24393
24394
24395
24396
24397
24398
24399
24400
24401
24402
24403
24404
24405
24406
24407
24408
24409
24410
24411
24412
24413
24414
24415
24416
24417
24418
24419
24420
24421
24422
24423
24424
24425
24426
24427
24428
24429
24430
24431
24432
24433
24434
24435
24436
24437
24438
24439
24440
24441
24442
24443
24444
24445
24446
24447
24448
24449
24450
24451
24452
24453
24454
24455
24456
24457
24458
24459
24460
24461
24462
24463
24464
24465
24466
24467
24468
24469
24470
24471
24472
24473
24474
24475
24476
24477
24478
24479
24480
24481
24482
24483
24484
24485
24486
24487
24488
24489
24490
24491
24492
24493
24494
24495
24496
24497
24498
24499
24500
24501
24502
24503
24504
24505
24506
24507
24508
24509
24510
24511
24512
24513
24514
24515
24516
24517
24518
24519
24520
24521
24522
24523
24524
24525
24526
24527
24528
24529
24530
24531
24532
24533
24534
24535
24536
24537
24538
24539
24540
24541
24542
24543
24544
24545
24546
24547
24548
24549
24550
24551
24552
24553
24554
24555
24556
24557
24558
24559
24560
24561
24562
24563
24564
24565
24566
24567
24568
24569
24570
24571
24572
24573
24574
24575
24576
24577
24578
24579
24580
24581
24582
24583
24584
24585
24586
24587
24588
24589
24590
24591
24592
24593
24594
24595
24596
24597
24598
24599
24600
24601
24602
24603
24604
24605
24606
24607
24608
24609
24610
24611
24612
24613
24614
24615
24616
24617
24618
24619
24620
24621
24622
24623
24624
24625
24626
24627
24628
24629
24630
24631
24632
24633
24634
24635
24636
24637
24638
24639
24640
24641
24642
24643
24644
24645
24646
24647
24648
24649
24650
24651
24652
24653
24654
24655
24656
24657
24658
24659
24660
24661
24662
24663
24664
24665
24666
24667
24668
24669
24670
24671
24672
24673
24674
24675
24676
24677
24678
24679
24680
24681
24682
24683
24684
24685
24686
24687
24688
24689
24690
24691
24692
24693
24694
24695
24696
24697
24698
24699
24700
24701
24702
24703
24704
24705
24706
24707
24708
24709
24710
24711
24712
24713
24714
24715
24716
24717
24718
24719
24720
24721
24722
24723
24724
24725
24726
24727
24728
24729
24730
24731
24732
24733
24734
24735
24736
24737
24738
24739
24740
24741
24742
24743
24744
24745
24746
24747
24748
24749
24750
24751
24752
24753
24754
24755
24756
24757
24758
24759
24760
24761
24762
24763
24764
24765
24766
24767
24768
24769
24770
24771
24772
24773
24774
24775
24776
24777
24778
24779
24780
24781
24782
24783
24784
24785
24786
24787
24788
24789
24790
24791
24792
24793
24794
24795
24796
24797
24798
24799
24800
24801
24802
24803
24804
24805
24806
24807
24808
24809
24810
24811
24812
24813
24814
24815
24816
24817
24818
24819
24820
24821
24822
24823
24824
24825
24826
24827
24828
24829
24830
24831
24832
24833
24834
24835
24836
24837
24838
24839
24840
24841
24842
24843
24844
24845
24846
24847
24848
24849
24850
24851
24852
24853
24854
24855
24856
24857
24858
24859
24860
24861
24862
24863
24864
24865
24866
24867
24868
24869
24870
24871
24872
24873
24874
24875
24876
24877
24878
24879
24880
24881
24882
24883
24884
24885
24886
24887
24888
24889
24890
24891
24892
24893
24894
24895
24896
24897
24898
24899
24900
24901
24902
24903
24904
24905
24906
24907
24908
24909
24910
24911
24912
24913
24914
24915
24916
24917
24918
24919
24920
24921
24922
24923
24924
24925
24926
24927
24928
24929
24930
24931
24932
24933
24934
24935
24936
24937
24938
24939
24940
24941
24942
24943
24944
24945
24946
24947
24948
24949
24950
24951
24952
24953
24954
24955
24956
24957
24958
24959
24960
24961
24962
24963
24964
24965
24966
24967
24968
24969
24970
24971
24972
24973
24974
24975
24976
24977
24978
24979
24980
24981
24982
24983
24984
24985
24986
24987
24988
24989
24990
24991
24992
24993
24994
24995
24996
24997
24998
24999
25000
25001
25002
25003
25004
25005
25006
25007
25008
25009
25010
25011
25012
25013
25014
25015
25016
25017
25018
25019
25020
25021
25022
25023
25024
25025
25026
25027
25028
25029
25030
25031
25032
25033
25034
25035
25036
25037
25038
25039
25040
25041
25042
25043
25044
25045
25046
25047
25048
25049
25050
25051
25052
25053
25054
25055
25056
25057
25058
25059
25060
25061
25062
25063
25064
25065
25066
25067
25068
25069
25070
25071
25072
25073
25074
25075
25076
25077
25078
25079
25080
25081
25082
25083
25084
25085
25086
25087
25088
25089
25090
25091
25092
25093
25094
25095
25096
25097
25098
25099
25100
25101
25102
25103
25104
25105
25106
25107
25108
25109
25110
25111
25112
25113
25114
25115
25116
25117
25118
25119
25120
25121
25122
25123
25124
25125
25126
25127
25128
25129
25130
25131
25132
25133
25134
25135
25136
25137
25138
25139
25140
25141
25142
25143
25144
25145
25146
25147
25148
25149
25150
25151
25152
25153
25154
25155
25156
25157
25158
25159
25160
25161
25162
25163
25164
25165
25166
25167
25168
25169
25170
25171
25172
25173
25174
25175
25176
25177
25178
25179
25180
25181
25182
25183
25184
25185
25186
25187
25188
25189
25190
25191
25192
25193
25194
25195
25196
25197
25198
25199
25200
25201
25202
25203
25204
25205
25206
25207
25208
25209
25210
25211
25212
25213
25214
25215
25216
25217
25218
25219
25220
25221
25222
25223
25224
25225
25226
25227
25228
25229
25230
25231
25232
25233
25234
25235
25236
25237
25238
25239
25240
25241
25242
25243
25244
25245
25246
25247
25248
25249
25250
25251
25252
25253
25254
25255
25256
25257
25258
25259
25260
25261
25262
25263
25264
25265
25266
25267
25268
25269
25270
25271
25272
25273
25274
25275
25276
25277
25278
25279
25280
25281
25282
25283
25284
25285
25286
25287
25288
25289
25290
25291
25292
25293
25294
25295
25296
25297
25298
25299
25300
25301
25302
25303
25304
25305
25306
25307
25308
25309
25310
25311
25312
25313
25314
25315
25316
25317
25318
25319
25320
25321
25322
25323
25324
25325
25326
25327
25328
25329
25330
25331
25332
25333
25334
25335
25336
25337
25338
25339
25340
25341
25342
25343
25344
25345
25346
25347
25348
25349
25350
25351
25352
25353
25354
25355
25356
25357
25358
25359
25360
25361
25362
25363
25364
25365
25366
25367
25368
25369
25370
25371
25372
25373
25374
25375
25376
25377
25378
25379
25380
25381
25382
25383
25384
25385
25386
25387
25388
25389
25390
25391
25392
25393
25394
25395
25396
25397
25398
25399
25400
25401
25402
25403
25404
25405
25406
25407
25408
25409
25410
25411
25412
25413
25414
25415
25416
25417
25418
25419
25420
25421
25422
25423
25424
25425
25426
25427
25428
25429
25430
25431
25432
25433
25434
25435
25436
25437
25438
25439
25440
25441
25442
25443
25444
25445
25446
25447
25448
25449
25450
25451
25452
25453
25454
25455
25456
25457
25458
25459
25460
25461
25462
25463
25464
25465
25466
25467
25468
25469
25470
25471
25472
25473
25474
25475
25476
25477
25478
25479
25480
25481
25482
25483
25484
25485
25486
25487
25488
25489
25490
25491
25492
25493
25494
25495
25496
25497
25498
25499
25500
25501
25502
25503
25504
25505
25506
25507
25508
25509
25510
25511
25512
25513
25514
25515
25516
25517
25518
25519
25520
25521
25522
25523
25524
25525
25526
25527
25528
25529
25530
25531
25532
25533
25534
25535
25536
25537
25538
25539
25540
25541
25542
25543
25544
25545
25546
25547
25548
25549
25550
25551
25552
25553
25554
25555
25556
25557
25558
25559
25560
25561
25562
25563
25564
25565
25566
25567
25568
25569
25570
25571
25572
25573
25574
25575
25576
25577
25578
25579
25580
25581
25582
25583
25584
25585
25586
25587
25588
25589
25590
25591
25592
25593
25594
25595
25596
25597
25598
25599
25600
25601
25602
25603
25604
25605
25606
25607
25608
25609
25610
25611
25612
25613
25614
25615
25616
25617
25618
25619
25620
25621
25622
25623
25624
25625
25626
25627
25628
25629
25630
25631
25632
25633
25634
25635
25636
25637
25638
25639
25640
25641
25642
25643
25644
25645
25646
25647
25648
25649
25650
25651
25652
25653
25654
25655
25656
25657
25658
25659
25660
25661
25662
25663
25664
25665
25666
25667
25668
25669
25670
25671
25672
25673
25674
25675
25676
25677
25678
25679
25680
25681
25682
25683
25684
25685
25686
25687
25688
25689
25690
25691
25692
25693
25694
25695
25696
25697
25698
25699
25700
25701
25702
25703
25704
25705
25706
25707
25708
25709
25710
25711
25712
25713
25714
25715
25716
25717
25718
25719
25720
25721
25722
25723
25724
25725
25726
25727
25728
25729
25730
25731
25732
25733
25734
25735
25736
25737
25738
25739
25740
25741
25742
25743
25744
25745
25746
25747
25748
25749
25750
25751
25752
25753
25754
25755
25756
25757
25758
25759
25760
25761
25762
25763
25764
25765
25766
25767
25768
25769
25770
25771
25772
25773
25774
25775
25776
25777
25778
25779
25780
25781
25782
25783
25784
25785
25786
25787
25788
25789
25790
25791
25792
25793
25794
25795
25796
25797
25798
25799
25800
25801
25802
25803
25804
25805
25806
25807
25808
25809
25810
25811
25812
25813
25814
25815
25816
25817
25818
25819
25820
25821
25822
25823
25824
25825
25826
25827
25828
25829
25830
25831
25832
25833
25834
25835
25836
25837
25838
25839
25840
25841
25842
25843
25844
25845
25846
25847
25848
25849
25850
25851
25852
25853
25854
25855
25856
25857
25858
25859
25860
25861
25862
25863
25864
25865
25866
25867
25868
25869
25870
25871
25872
25873
25874
25875
25876
25877
25878
25879
25880
25881
25882
25883
25884
25885
25886
25887
25888
25889
25890
25891
25892
25893
25894
25895
25896
25897
25898
25899
25900
25901
25902
25903
25904
25905
25906
25907
25908
25909
25910
25911
25912
25913
25914
25915
25916
25917
25918
25919
25920
25921
25922
25923
25924
25925
25926
25927
25928
25929
25930
25931
25932
25933
25934
25935
25936
25937
25938
25939
25940
25941
25942
25943
25944
25945
25946
25947
25948
25949
25950
25951
25952
25953
25954
25955
25956
25957
25958
25959
25960
25961
25962
25963
25964
25965
25966
25967
25968
25969
25970
25971
25972
25973
25974
25975
25976
25977
25978
25979
25980
25981
25982
25983
25984
25985
25986
25987
25988
25989
25990
25991
25992
25993
25994
25995
25996
25997
25998
25999
26000
26001
26002
26003
26004
26005
26006
26007
26008
26009
26010
26011
26012
26013
26014
26015
26016
26017
26018
26019
26020
26021
26022
26023
26024
26025
26026
26027
26028
26029
26030
26031
26032
26033
26034
26035
26036
26037
26038
26039
26040
26041
26042
26043
26044
26045
26046
26047
26048
26049
26050
26051
26052
26053
26054
26055
26056
26057
26058
26059
26060
26061
26062
26063
26064
26065
26066
26067
26068
26069
26070
26071
26072
26073
26074
26075
26076
26077
26078
26079
26080
26081
26082
26083
26084
26085
26086
26087
26088
26089
26090
26091
26092
26093
26094
26095
26096
26097
26098
26099
26100
26101
26102
26103
26104
26105
26106
26107
26108
26109
26110
26111
26112
26113
26114
26115
26116
26117
26118
26119
26120
26121
26122
26123
26124
26125
26126
26127
26128
26129
26130
26131
26132
26133
26134
26135
26136
26137
26138
26139
26140
26141
26142
26143
26144
26145
26146
26147
26148
26149
26150
26151
26152
26153
26154
26155
26156
26157
26158
26159
26160
26161
26162
26163
26164
26165
26166
26167
26168
26169
26170
26171
26172
26173
26174
26175
26176
26177
26178
26179
26180
26181
26182
26183
26184
26185
26186
26187
26188
26189
26190
26191
26192
26193
26194
26195
26196
26197
26198
26199
26200
26201
26202
26203
26204
26205
26206
26207
26208
26209
26210
26211
26212
26213
26214
26215
26216
26217
26218
26219
26220
26221
26222
26223
26224
26225
26226
26227
26228
26229
26230
26231
26232
26233
26234
26235
26236
26237
26238
26239
26240
26241
26242
26243
26244
26245
26246
26247
26248
26249
26250
26251
26252
26253
26254
26255
26256
26257
26258
26259
26260
26261
26262
26263
26264
26265
26266
26267
26268
26269
26270
26271
26272
26273
26274
26275
26276
26277
26278
26279
26280
26281
26282
26283
26284
26285
26286
26287
26288
26289
26290
26291
26292
26293
26294
26295
26296
26297
26298
26299
26300
26301
26302
26303
26304
26305
26306
26307
26308
26309
26310
26311
26312
26313
26314
26315
26316
26317
26318
26319
26320
26321
26322
26323
26324
26325
26326
26327
26328
26329
26330
26331
26332
26333
26334
26335
26336
26337
26338
26339
26340
26341
26342
26343
26344
26345
26346
26347
26348
26349
26350
26351
26352
26353
26354
26355
26356
26357
26358
26359
26360
26361
26362
26363
26364
26365
26366
26367
26368
26369
26370
26371
26372
26373
26374
26375
26376
26377
26378
26379
26380
26381
26382
26383
26384
26385
26386
26387
26388
26389
26390
26391
26392
26393
26394
26395
26396
26397
26398
26399
26400
26401
26402
26403
26404
26405
26406
26407
26408
26409
26410
26411
26412
26413
26414
26415
26416
26417
26418
26419
26420
26421
26422
26423
26424
26425
26426
26427
26428
26429
26430
26431
26432
26433
26434
26435
26436
26437
26438
26439
26440
26441
26442
26443
26444
26445
26446
26447
26448
26449
26450
26451
26452
26453
26454
26455
26456
26457
26458
26459
26460
26461
26462
26463
26464
26465
26466
26467
26468
26469
26470
26471
26472
26473
26474
26475
26476
26477
26478
26479
26480
26481
26482
26483
26484
26485
26486
26487
26488
26489
26490
26491
26492
26493
26494
26495
26496
26497
26498
26499
26500
26501
26502
26503
26504
26505
26506
26507
26508
26509
26510
26511
26512
26513
26514
26515
26516
26517
26518
26519
26520
26521
26522
26523
26524
26525
26526
26527
26528
26529
26530
26531
26532
26533
26534
26535
26536
26537
26538
26539
26540
26541
26542
26543
26544
26545
26546
26547
26548
26549
26550
26551
26552
26553
26554
26555
26556
26557
26558
26559
26560
26561
26562
26563
26564
26565
26566
26567
26568
26569
26570
26571
26572
26573
26574
26575
26576
26577
26578
26579
26580
26581
26582
26583
26584
26585
26586
26587
26588
26589
26590
26591
26592
26593
26594
26595
26596
26597
26598
26599
26600
26601
26602
26603
26604
26605
26606
26607
26608
26609
26610
26611
26612
26613
26614
26615
26616
26617
26618
26619
26620
26621
26622
26623
26624
26625
26626
26627
26628
26629
26630
26631
26632
26633
26634
26635
26636
26637
26638
26639
26640
26641
26642
26643
26644
26645
26646
26647
26648
26649
26650
26651
26652
26653
26654
26655
26656
26657
26658
26659
26660
26661
26662
26663
26664
26665
26666
26667
26668
26669
26670
26671
26672
26673
26674
26675
26676
26677
26678
26679
26680
26681
26682
26683
26684
26685
26686
26687
26688
26689
26690
26691
26692
26693
26694
26695
26696
26697
26698
26699
26700
26701
26702
26703
26704
26705
26706
26707
26708
26709
26710
26711
26712
26713
26714
26715
26716
26717
26718
26719
26720
26721
26722
26723
26724
26725
26726
26727
26728
26729
26730
26731
26732
26733
26734
26735
26736
26737
26738
26739
26740
26741
26742
26743
26744
26745
26746
26747
26748
26749
26750
26751
26752
26753
26754
26755
26756
26757
26758
26759
26760
26761
26762
26763
26764
26765
26766
26767
26768
26769
26770
26771
26772
26773
26774
26775
26776
26777
26778
26779
26780
26781
26782
26783
26784
26785
26786
26787
26788
26789
26790
26791
26792
26793
26794
26795
26796
26797
26798
26799
26800
26801
26802
26803
26804
26805
26806
26807
26808
26809
26810
26811
26812
26813
26814
26815
26816
26817
26818
26819
26820
26821
26822
26823
26824
26825
26826
26827
26828
26829
26830
26831
26832
26833
26834
26835
26836
26837
26838
26839
26840
26841
26842
26843
26844
26845
26846
26847
26848
26849
26850
26851
26852
26853
26854
26855
26856
26857
26858
26859
26860
26861
26862
26863
26864
26865
26866
26867
26868
26869
26870
26871
26872
26873
26874
26875
26876
26877
26878
26879
26880
26881
26882
26883
26884
26885
26886
26887
26888
26889
26890
26891
26892
26893
26894
26895
26896
26897
26898
26899
26900
26901
26902
26903
26904
26905
26906
26907
26908
26909
26910
26911
26912
26913
26914
26915
26916
26917
26918
26919
26920
26921
26922
26923
26924
26925
26926
26927
26928
26929
26930
26931
26932
26933
26934
26935
26936
26937
26938
26939
26940
26941
26942
26943
26944
26945
26946
26947
26948
26949
26950
26951
26952
26953
26954
26955
26956
26957
26958
26959
26960
26961
26962
26963
26964
26965
26966
26967
26968
26969
26970
26971
26972
26973
26974
26975
26976
26977
26978
26979
26980
26981
26982
26983
26984
26985
26986
26987
26988
26989
26990
26991
26992
26993
26994
26995
26996
26997
26998
26999
27000
27001
27002
27003
27004
27005
27006
27007
27008
27009
27010
27011
27012
27013
27014
27015
27016
27017
27018
27019
27020
27021
27022
27023
27024
27025
27026
27027
27028
27029
27030
27031
27032
27033
27034
27035
27036
27037
27038
27039
27040
27041
27042
27043
27044
27045
27046
27047
27048
27049
27050
27051
27052
27053
27054
27055
27056
27057
27058
27059
27060
27061
27062
27063
27064
27065
27066
27067
27068
27069
27070
27071
27072
27073
27074
27075
27076
27077
27078
27079
27080
27081
27082
27083
27084
27085
27086
27087
27088
27089
27090
27091
27092
27093
27094
27095
27096
27097
27098
27099
27100
27101
27102
27103
27104
27105
27106
27107
27108
27109
27110
27111
27112
27113
27114
27115
27116
27117
27118
27119
27120
27121
27122
27123
27124
27125
27126
27127
27128
27129
27130
27131
27132
27133
27134
27135
27136
27137
27138
27139
27140
27141
27142
27143
27144
27145
27146
27147
27148
27149
27150
27151
27152
27153
27154
27155
27156
27157
27158
27159
27160
27161
27162
27163
27164
27165
27166
27167
27168
27169
27170
27171
27172
27173
27174
27175
27176
27177
27178
27179
27180
27181
27182
27183
27184
27185
27186
27187
27188
27189
27190
27191
27192
27193
27194
27195
27196
27197
27198
27199
27200
27201
27202
27203
27204
27205
27206
27207
27208
27209
27210
27211
27212
27213
27214
27215
27216
27217
27218
27219
27220
27221
27222
27223
27224
27225
27226
27227
27228
27229
27230
27231
27232
27233
27234
27235
27236
27237
27238
27239
27240
27241
27242
27243
27244
27245
27246
27247
27248
27249
27250
27251
27252
27253
27254
27255
27256
27257
27258
27259
27260
27261
27262
27263
27264
27265
27266
27267
27268
27269
27270
27271
27272
27273
27274
27275
27276
27277
27278
27279
27280
27281
27282
27283
27284
27285
27286
27287
27288
27289
27290
27291
27292
27293
27294
27295
27296
27297
27298
27299
27300
27301
27302
27303
27304
27305
27306
27307
27308
27309
27310
27311
27312
27313
27314
27315
27316
27317
27318
27319
27320
27321
27322
27323
27324
27325
27326
27327
27328
27329
27330
27331
27332
27333
27334
27335
27336
27337
27338
27339
27340
27341
27342
27343
27344
27345
27346
27347
27348
27349
27350
27351
27352
27353
27354
27355
27356
27357
27358
27359
27360
27361
27362
27363
27364
27365
27366
27367
27368
27369
27370
27371
27372
27373
27374
27375
27376
27377
27378
27379
27380
27381
27382
27383
27384
27385
27386
27387
27388
27389
27390
27391
27392
27393
27394
27395
27396
27397
27398
27399
27400
27401
27402
27403
27404
27405
27406
27407
27408
27409
27410
27411
27412
27413
27414
27415
27416
27417
27418
27419
27420
27421
27422
27423
27424
27425
27426
27427
27428
27429
27430
27431
27432
27433
27434
27435
27436
27437
27438
27439
27440
27441
27442
27443
27444
27445
27446
27447
27448
27449
27450
27451
27452
27453
27454
27455
27456
27457
27458
27459
27460
27461
27462
27463
27464
27465
27466
27467
27468
27469
27470
27471
27472
27473
27474
27475
27476
27477
27478
27479
27480
27481
27482
27483
27484
27485
27486
27487
27488
27489
27490
27491
27492
27493
27494
27495
27496
27497
27498
27499
27500
27501
27502
27503
27504
27505
27506
27507
27508
27509
27510
27511
27512
27513
27514
27515
27516
27517
27518
27519
27520
27521
27522
27523
27524
27525
27526
27527
27528
27529
27530
27531
27532
27533
27534
27535
27536
27537
27538
27539
27540
27541
27542
27543
27544
27545
27546
27547
27548
27549
27550
27551
27552
27553
27554
27555
27556
27557
27558
27559
27560
27561
27562
27563
27564
27565
27566
27567
27568
27569
27570
27571
27572
27573
27574
27575
27576
27577
27578
27579
27580
27581
27582
27583
27584
27585
27586
27587
27588
27589
27590
27591
27592
27593
27594
27595
27596
27597
27598
27599
27600
27601
27602
27603
27604
27605
27606
27607
27608
27609
27610
27611
27612
27613
27614
27615
27616
27617
27618
27619
27620
27621
27622
27623
27624
27625
27626
27627
27628
27629
27630
27631
27632
27633
27634
27635
27636
27637
27638
27639
27640
27641
27642
27643
27644
27645
27646
27647
27648
27649
27650
27651
27652
27653
27654
27655
27656
27657
27658
27659
27660
27661
27662
27663
27664
27665
27666
27667
27668
27669
27670
27671
27672
27673
27674
27675
27676
27677
27678
27679
27680
27681
27682
27683
27684
27685
27686
27687
27688
27689
27690
27691
27692
27693
27694
27695
27696
27697
27698
27699
27700
27701
27702
27703
27704
27705
27706
27707
27708
27709
27710
27711
27712
27713
27714
27715
27716
27717
27718
27719
27720
27721
27722
27723
27724
27725
27726
27727
27728
27729
27730
27731
27732
27733
27734
27735
27736
27737
27738
27739
27740
27741
27742
27743
27744
27745
27746
27747
27748
27749
27750
27751
27752
27753
27754
27755
27756
27757
27758
27759
27760
27761
27762
27763
27764
27765
27766
27767
27768
27769
27770
27771
27772
27773
27774
27775
27776
27777
27778
27779
27780
27781
27782
27783
27784
27785
27786
27787
27788
27789
27790
27791
27792
27793
27794
27795
27796
27797
27798
27799
27800
27801
27802
27803
27804
27805
27806
27807
27808
27809
27810
27811
27812
27813
27814
27815
27816
27817
27818
27819
27820
27821
27822
27823
27824
27825
27826
27827
27828
27829
27830
27831
27832
27833
27834
27835
27836
27837
27838
27839
27840
27841
27842
27843
27844
27845
27846
27847
27848
27849
27850
27851
27852
27853
27854
27855
27856
27857
27858
27859
27860
27861
27862
27863
27864
27865
27866
27867
27868
27869
27870
27871
27872
27873
27874
27875
27876
27877
27878
27879
27880
27881
27882
27883
27884
27885
27886
27887
27888
27889
27890
27891
27892
27893
27894
27895
27896
27897
27898
27899
27900
27901
27902
27903
27904
27905
27906
27907
27908
27909
27910
27911
27912
27913
27914
27915
27916
27917
27918
27919
27920
27921
27922
27923
27924
27925
27926
27927
27928
27929
27930
27931
27932
27933
27934
27935
27936
27937
27938
27939
27940
27941
27942
27943
27944
27945
27946
27947
27948
27949
27950
27951
27952
27953
27954
27955
27956
27957
27958
27959
27960
27961
27962
27963
27964
27965
27966
27967
27968
27969
27970
27971
27972
27973
27974
27975
27976
27977
27978
27979
27980
27981
27982
27983
27984
27985
27986
27987
27988
27989
27990
27991
27992
27993
27994
27995
27996
27997
27998
27999
28000
28001
28002
28003
28004
28005
28006
28007
28008
28009
28010
28011
28012
28013
28014
28015
28016
28017
28018
28019
28020
28021
28022
28023
28024
28025
28026
28027
28028
28029
28030
28031
28032
28033
28034
28035
28036
28037
28038
28039
28040
28041
28042
28043
28044
28045
28046
28047
28048
28049
28050
28051
28052
28053
28054
28055
28056
28057
28058
28059
28060
28061
28062
28063
28064
28065
28066
28067
28068
28069
28070
28071
28072
28073
28074
28075
28076
28077
28078
28079
28080
28081
28082
28083
28084
28085
28086
28087
28088
28089
28090
28091
28092
28093
28094
28095
28096
28097
28098
28099
28100
28101
28102
28103
28104
28105
28106
28107
28108
28109
28110
28111
28112
28113
28114
28115
28116
28117
28118
28119
28120
28121
28122
28123
28124
28125
28126
28127
28128
28129
28130
28131
28132
28133
28134
28135
28136
28137
28138
28139
28140
28141
28142
28143
28144
28145
28146
28147
28148
28149
28150
28151
28152
28153
28154
28155
28156
28157
28158
28159
28160
28161
28162
28163
28164
28165
28166
28167
28168
28169
28170
28171
28172
28173
28174
28175
% This manual is copyright (C) 1984 by the American Mathematical Society.
% All rights are reserved!
% The file is distributed only for people to see its examples of TeX input,
% not for use in the preparation of books like The TeXbook.
% Permission for any other use of this file must be obtained in writing
% from the copyright holder and also from the publisher (Addison-Wesley).
\loop\iftrue
  \errmessage{This manual is copyrighted and should not be TeXed}\repeat
\pausing1 \input manmac
\ifproofmode\message{Proof mode is on!}\pausing1\fi
% halftitle
\titlepage
\pageno=-1983
\null\bigskip
\line{\cmman The\hfill T\kern-10pt\lower13pt\hbox{E}\kern-5pt Xbook}
\vfill
\ifproofmode
\rightline{The fine print in the upper right-hand}
\rightline{corner of each page is a draft of intended}
\rightline{index entries; it won't appear in the real book.}
\rightline{Some index entries will be in |typewriter type|}
\rightline{and/or preceded by {\tt\char`\\} or enclosed in \<$\ldots$>, etc;}
\rightline{such typographic distinctions aren't shown here.}
\rightline{An index entry often extends for several pages;}
\rightline{the actual scope will be determined later.}
\rightline{Please note things that should be indexed but aren't.}
\fi
\eject
\titlepage\null\vfill\eject % blank page
% title
\pageno=-1 % the front matter is numbered with roman numerals
\font\auth=cmssdc10 scaled\magstep4 % used only on the title page
\font\elevenbf=cmbx10 scaled\magstephalf % ditto
\font\elevenit=cmti10 scaled\magstephalf % ditto
\font\elevenrm=cmr10 scaled\magstephalf % ditto
\titlepage
\null\bigskip
\line{\cmman The\hfill T\kern-10pt\lower13pt\hbox{E}\kern-5pt Xbook}
^^{Knuth, Donald Ervin}
^^{Bibby, Duane Robert}
\vskip 1pc
\baselineskip 13pt \elevenbf
\halign to\hsize{#\hfil\tabskip 0pt plus 1fil&#\hfil\tabskip0pt\cr
\kern5.5mm\auth DONALD \kern-1pt E. \kern-1pt KNUTH&
  \elevenit Stanford University\cr
\noalign{\vskip 12pc}
&\elevenit I\kern.7ptllustrations by\cr
&DU\kern-1ptANE BIBBY\cr
\noalign{\vfill}
&\setbox0=\hbox{\manual77}%
\setbox2=\hbox to\wd0{\hss\manual6\hss}%
\raise2.3mm\box2\kern-\wd0\box0\cr % A-W logo
&ADDISON\kern.1em--WESLEY\cr
%&PUBLISHING COMP\kern-.13emANY\kern-1.5mm\cr
\noalign{\vskip.5pc \global\elevenrm}
&Upper Saddle River, NJ\cr
&Boston\enspace$\cdot$\enspace Indianapolis\cr
&San Francisco\enspace$\cdot$\enspace New York\cr
&Toronto\enspace$\cdot$\enspace Montr\'eal\cr
&London\enspace$\cdot$\enspace Munich\cr
&Paris\enspace$\cdot$\enspace Madrid\cr
&Capetown\enspace$\cdot$\enspace Sydney\enspace$\cdot$\enspace Tokyo\cr
&Singapore\enspace$\cdot$\enspace Mexico City\cr}
\kern24pt
\eject
% copyright
\titlepage
\eightpoint
\vbox to 8pc{}
\noindent\strut
This manual describes \TeX\ Version 3.0. Some
of the advanced features mentioned here are absent from earlier versions.
\medskip
\noindent
The quotation on page \sesame\ is copyright $\copyright$ 1970 by Sesame
Street, Inc., and used by permission of the Children's Television Workshop.
\medskip
\noindent
\TeX\ is a trademark of the American Mathematical Society.
\medskip
\noindent
{\manual opqrstuq} is a trademark of Addison\kern.1em--Wesley
 Publishing Company.
\bigskip\medskip
\noindent
{\bf Library of Congress cataloging in publication data}
\medskip
{\tt\halign{#\hfil\cr
Knuth, Donald Ervin, 1938-\cr
\ \ \  The TeXbook.\cr
\noalign{\medskip}
\ \ \  (Computers \& Typesetting ; A)\cr
\ \ \  Includes index.\cr
\ \ \  1.~TeX (Computer system).\ \ 2.~Computerized\cr
typesetting.\ \ 3.~Mathematics printing.\ \ I.~Title.\cr
II.\ Series:\ \ Knuth, Donald Ervin, 1938-\ \ \ \ .\cr
Computers \& typesetting ; A.\cr
Z253.4.T47K58\ \ 1986\ \ \ \ \ \ \ \ \ 686.2\char13 2544\ \ \ \ \ \ 85-30845\cr
ISBN 0-201-13447-0\cr
ISBN 0-201-13448-9 (soft)\cr}}
\vfill
\noindent
%{\sl \kern-1pt First hardcover edition, January 1986}
%{\sl \kern-1pt Incorporates the final corrections made in 1996, and a few dozen more.}
%{\sl \kern-1pt Incorporates all corrections known in 2013.}
{\sl \kern-1pt Incorporates all corrections known in 2020.}
\smallskip
\noindent
Internet page {\tt http://www-cs-faculty.stanford.edu/\char`\~
 knuth/abcde.html}
contains current information about this book and related books.
\smallskip
\noindent
Copyright $\copyright$ 1984, 1986 by the American Mathematical Society
\smallskip
\noindent
This book is published jointly by the American Mathematical Society
and Addison\kern.1em--Wesley Publishing Company.
All rights reserved.
%No part of this publication may be reproduced, stored in
%a retrieval system, or transmitted, in any form or by any means,
%electronic, mechanical, photocopying, recording, or otherwise, without
%the prior written permission of the publishers.
This publication is protected by copyright, and permission must be
obtained from the publisher prior to any prohibited reproduction, storage in
a~retrieval system, or transmission in any form or by any means, electronic,
mechanical, photocopying, recording, or likewise. For information regarding
permissions, request forms, and the appropriate contacts with the
Pearson Education Global Rights \& Permissions Department, please visit
 {\tt www.pearson.com/permissions/}.
Printed in the United States of America. % Published simultaneously in Canada.
\medskip
\noindent
%ISBN 0-201-13448-9\par % paperback
%ISBN 0-201-13447-0\par % hardcover
ISBN-13 \enspace 978-0-201-13447-6\par\noindent
ISBN-10 \enspace\phantom{978-}0-201-13447-0\par
%33 34 35 36 37 38 39 DOC 09 08 07 06 % paperback
%18 19 20 21 22 23 24 DOC 10 09 08 07 06 % hardcover
\smallskip\noindent
%Text printed in the United States
%% at Courier Westford in Westford, Massachusetts.\par\noindent
%at LSC Communications in Crawfordsville, Indiana.\par\noindent
%%Nineteenth Printing, February 2012
%% [The printer deleted "Text printed ... Crawfordsville..." in 2017!]
%Thirty-fifth Softcover Printing, April 2017\par\noindent
%35\quad17
%Twenty-second Printing, January 2021\par\noindent
Twenty-third Printing, February 2021\par\noindent
%20\quad17
\smallskip
\font\pearsonkluj=arial at 9pt
\leftline{\pearsonkluj ScoutAutomatedPrintCode}
^^{Knuth, Donald Ervin}
^^|\copyright|
\eject
% dedication
\titlepage
\vbox to 8pc{}
\rightline{\strut\eightssi To Jill:}
^^{Knuth, Jill Carter}
\vskip2pt
\rightline{\eightssi For your books and brochures}
\vfill
\eject
% blank page
\titlepage
\null\vfill
\eject
% the preface
\titlepage
\def\rhead{Preface}
\vbox to 8pc{
\rightline{\titlefont Preface}\vss}
{\topskip 9pc % this makes equal sinkage throughout the Preface
\vskip-\parskip
\tenpoint
\noindent\hang\hangafter-2
\smash{\lower12pt\hbox to 0pt{\hskip-\hangindent\cmman G\hfill}}\hskip-16pt
{\sc ENTLE} R{\sc EADER}: \strut This is a handbook about
\TeX, a new typesetting system intended for the creation
of beautiful books---and especially for books that contain a lot of
mathematics. By preparing a manuscript in \TeX\ format, you will be
telling a computer exactly how the manuscript is to be transformed into
pages whose typographic quality is comparable to that of the world's
finest printers; yet you won't need to do much more work than would be
involved if you were simply typing the manuscript on an ordinary
typewriter. In fact, your total work will probably be significantly less,
if you consider the time it ordinarily takes to revise a typewritten manuscript,
since computer text files are so easy to change and to reprocess. \
(If such claims sound too good to be true, keep in mind that they were made
by \TeX's designer, on a day when \TeX\ happened to
be working, so the statements may be biased; but read on anyway.)

This manual is intended for people who have never used \TeX\ before, as
well as for experienced \TeX\ hackers. In other words, it's supposed to
be a panacea that satisfies everybody, at the risk of satisfying nobody.
Everything you need to know about \TeX\ is explained
here somewhere, and so are a lot of things that most users don't care about.
If you are preparing a simple manuscript, you won't need to
learn much about \TeX\ at all; on the other hand, some
things that go into the printing of technical books are inherently
difficult, and if you wish to achieve more complex effects you
will want to penetrate some of \TeX's darker corners. In order
to make it possible for many types of users to read this manual
effectively, a special sign is used to designate material that is
for wizards only: When the symbol
$$\vbox{\hbox{\dbend}\vskip 11pt}$$
appears at the beginning of a paragraph, it warns of a ``^{dangerous bend}''
in the train of thought; don't read the paragraph unless you need to.
Brave and experienced drivers at the controls of \TeX\ will gradually enter
more and more of these hazardous areas, but for most applications the
details won't matter.

All that you really ought to know, before reading on, is how to get a
file of text into your computer using a standard editing program. This
manual explains what that file ought to look like so that \TeX\ will
understand it, but basic computer usage is not explained here.
Some previous experience with technical typing will be quite helpful
if you plan to do heavily mathematical work with \TeX, although it
is not absolutely necessary. \TeX\ will do most of the necessary
formatting of equations automatically; but users with more experience
will be able to obtain better results, since there are so many ways
to deal with formulas.

Some of the paragraphs in this manual are so esoteric that they are rated
$$\vcenter{\hbox{\dbend\kern1pt\dbend}\vskip 11pt}\;;$$
everything that was said about single dangerous-bend signs goes double
for these. You should probably have at least a month's experience with
\TeX\ before you attempt to fathom such doubly dangerous depths
of the system; in fact, most people will never need to know \TeX\
in this much detail, even if they use it every day. After all, it's
possible to drive a car without knowing how the engine works.
Yet the whole story is here in case you're curious. \ (About \TeX, not cars.)

The reason for such different levels of complexity is that people change
as they grow accustomed to any powerful tool. When you first try to use \TeX,
you'll find that some parts of it are very easy, while other things will take
some getting used to. A day or so later, after you have successfully typeset a
few pages, you'll be a different person; the concepts that used to bother you
will now seem natural, and you'll be able to picture the final result in
your mind before it comes out of the machine. But you'll probably run into
challenges of a different kind. After another week your perspective will
change again, and you'll grow in yet another way; and so on. As years go by,
you might become involved with many different kinds of typesetting; and
you'll find that your usage of \TeX\ will keep changing as your experience
builds. That's the way it is with any powerful tool: There's always more
to learn, and there are always better ways to do what you've done before.
At every stage in the development you'll want a slightly different sort of
manual.  You may even want to write one yourself.  By paying attention to
the dangerous bend signs in this book you'll be better able to focus on
the level that interests you at a particular time.

Computer system manuals usually make dull reading, but take heart:
This one contains {\sc ^{JOKES}} every once in a while, so you might actually
enjoy reading it. \ (However, most of the jokes can only be appreciated
properly if you understand a technical point that is being made---so
read {\sl carefully}.)

Another noteworthy characteristic of this manual is that it doesn't
always tell the ^{truth}. When certain concepts of \TeX\ are introduced
informally, general rules will be stated; afterwards you will find that the
rules aren't strictly true. In general, the later chapters contain more
reliable information than the earlier ones do. The author feels that this
technique of deliberate lying will actually make it easier for you to
learn the ideas. Once you understand a simple but false rule, it will not
be hard to supplement that rule with its exceptions.

In order to help you internalize what you're reading,
{\sc ^{EXERCISES}} are sprinkled through this manual. It is generally intended
that every reader should try every exercise, except for questions that appear
in the ``dangerous bend'' areas. If you can't solve a problem, you
can always look up the answer.
But please, try first to solve it by yourself; then you'll learn more
and you'll learn faster. Furthermore, if you think you do know the solution,
you should turn to Appendix~A and check it out, just to make sure.

The \TeX\ language described in this book is similar to the author's first
attempt at a document formatting language, but the new system differs
from the old~one in literally thousands of details. Both languages have
been called \TeX; but henceforth the old language should be called
\TeX78, and its use should rapidly fade away. Let's keep the name \TeX\
for the language described here, since it is so much better, and since
it is not going to change any more. ^^{TeX78}

I wish to thank the hundreds of people who have helped me to formulate
this ``definitive edition'' of the \TeX\ language, based on their
experiences with preliminary versions of the system. My work at Stanford
has been generously supported by the ^{National Science Foundation}, the
^{Office of Naval Research}, the ^{IBM Corporation}, and the ^{System
Development Foundation}. I also wish to thank the ^{American Mathematical
Society} for its encouragement, for establishing the \TeX\ Users Group,
and for publishing the {\sl ^{TUGboat}\/} newsletter (see Appendix~J).

\medskip
\line{{\sl Stanford, California}\hfil--- D. E. K.}^^{Knuth, Don}
\line{\sl June 1983\hfil}

} % end of the special \topskip
\endchapter

`Tis pleasant, sure, to see one's name in print;
A book's a book, although there's nothing in 't.
\author ^{BYRON}, {\sl English Bards and Scotch Reviewers\/} (1809)

\bigskip

A question arose as to whether we were covering the field
that it was intended we should fill with this manual.
\author RICHARD R. ^{DONNELLEY}, {\sl Proceedings, United %
  Typothet{\ae} of America\/} (1897)

\eject
% the table of contents
\titlepage
\vbox to 8pc{
\rightline{\titlefont Contents}
\vfill}
^^{Contents of this manual, table}
\def\rhead{Contents}
\tenpoint
\begingroup
\countdef\counter=255
\def\diamondleaders{\global\advance\counter by 1
  \ifodd\counter \kern-10pt \fi
  \leaders\hbox to 20pt{\ifodd\counter \kern13pt \else\kern3pt \fi
    .\hss}}
\baselineskip 15pt plus 5pt
\def\\#1. #2. #3.{\line{\strut
    \hbox to\parindent{\bf\hbox to 1em{\hss#1}\hss}%
    \rm#2\diamondleaders\hfil\hbox to 2em{\hss#3}}}
\\1. The Name of the Game. 1.
\\2. Book Printing versus Ordinary Typing. 3.
\\3. Controlling \TeX. 7.
\\4. Fonts of Type. 13.
\\5. Grouping. 19.
\\6. Running \TeX. 23.
\\7. How \TeX\ Reads What You Type. 37.
\\8. The Characters You Type. 43.
\\9. \TeX's Roman Fonts. 51.
\\10. Dimensions. 57.
\\11. Boxes. 63.
\\12. Glue. 69.
\\13. Modes. 85.
\\14. How \TeX\ Breaks Paragraphs into Lines. 91.
\\15. How \TeX\ Makes Lines into Pages. 109.
\\16. Typing Math Formulas. 127.
\\17. More about Math. 139.
\\18. Fine Points of Mathematics Typing. 161.
\\19. Displayed Equations. 185.
\\20. Definitions (also called Macros). 199.
\\21. Making Boxes. 221.
\\22. Alignment. 231.
\\23. Output Routines. 251.
\eject
\vbox to 8pc{}
\\24. Summary of Vertical Mode. 267.
\\25. Summary of Horizontal Mode. 285.
\\26. Summary of Math Mode. 289.
\\27. Recovery from Errors. 295.
\null
\leftline{\indent\bf Appendices}
\\A. Answers to All the Exercises. 305.
\\B. Basic Control Sequences. 339.
\\C. Character Codes. 367.
\\D. Dirty Tricks. 373.
\\E. Example Formats. 403.
\\F. Font Tables. 427.
\\G. Generating Boxes from Formulas. 441.
\\H. Hyphenation. 449.
\\I\hskip 1pt. Index. 457.
\\J\hskip 1pt. Joining the \TeX\ Community. 483.
\null % 17 lines so far to balance the 23 on the other page
\null % 18
\null % 19
\null % 20
\null % 21
\null % 22
\null % 23
\eject
\endgroup
\beginchapter Chapter 1. The Name of\\the Game

\pageno=1 % This is page number 1, number 1,
English words like `technology' stem from a Greek root beginning with
the letters $\tau\epsilon\chi\ldots\,$; and this same Greek word means {\sl
art\/} as well as technology. Hence the name \TeX, which is an
uppercase form of $\tau\epsilon\chi$.^^{TeX (actually \TeX), meaning of}
^^|\tau|^^|\epsilon|^^|\chi|

Insiders pronounce the $\chi$ of \TeX\ as a Greek chi, not as an `x', so that
\TeX\ rhymes with the word blecchhh. It's the `ch' sound in Scottish words
like {\sl loch\/} or German words like {\sl ach\/}; it's a Spanish `j' and a
Russian `kh'. When you say it correctly to your computer, the terminal
may become slightly moist.

The purpose of this pronunciation exercise is to remind you that \TeX\ is
primarily concerned with high-quality technical manuscripts: Its emphasis is
on art and technology, as in the underlying Greek word. If you merely want
to produce a passably good document---something acceptable and basically
readable but not really beautiful---a simpler system will usually suffice.
With \TeX\ the goal is to produce the {\sl finest\/} quality; this requires
more attention to detail, but you will not find it much harder to go the
extra distance, and you'll be able to take special pride in the finished
product. ^^{beauty} % since truth is indexed, ...

On the other hand, it's important to notice another thing about \TeX's name:
The `E' is out of kilter. This ^^{logo}
displaced `E' is a reminder that \TeX\ is about typesetting, and it
distinguishes \TeX\ from other system names. In fact, ^{TEX} (pronounced
{\sl tecks\/}) is the admirable {\sl Text EXecutive\/} processor developed by
^{Honeywell Information Systems}. Since these two system names are
^^{Bemer, Robert, see TEX, ASCII}
pronounced quite differently, they should also be spelled differently. The
correct way to refer to \TeX\ in a computer file, or when using some other
medium that doesn't allow lowering of the `E', is to type `^|TeX|'. Then
there will be no confusion with similar names, and people will be
primed to pronounce everything properly.

\exercise After you have mastered the material in this book, what will
you be: a \TeX pert, or a \TeX nician?
\answer A \TeX nician (underpaid); sometimes also called a \TeX acker.

\endchapter

They do certainly give
very strange and new-fangled names to diseases.
\author ^{PLATO}, {\sl The Republic}, Book 3 (c.\ 375 B.C.) % 405c

\bigskip

Technique! The very word is like the shriek
Of outraged Art. It is the idiot name
Given to effort by those who are too weak,
Too weary, or too dull to play the game.
\author LEONARD ^{BACON}, {\sl Sophia Trenton\/} (1920) % composed at Stanford

\eject
\beginchapter Chapter 2. Book Printing\\versus\\Ordinary Typing

When you first started using a computer terminal, you probably had to adjust
to the difference between the digit `1' and the lowercase letter `l'.
When you take the next step to the level of typography that is common in
book publishing, a few more adjustments of the same kind need to be made;
your eyes and your fingers need to learn to make a few more distinctions.

In the first place, there are two kinds of ^{quotation marks} in books, but
only one kind on the typewriter. Even your computer terminal, which has
more characters than an ordinary typewriter, probably has only a
non-oriented double-quote mark (|"|), because the standard ^{ASCII} code
for computers was not invented with book publishing in mind. However, your
terminal probably does have two flavors of single-quote marks, namely |`|
and |'|; the second of these is useful also as an ^{apostrophe}.
American keyboards usually contain a left-quote character that shows up
as something like {\tt\char'22}, and an apostrophe or right-quote that
looks like {\tt\char'15} or {\tt\char'23}.

To produce double-quote marks with \TeX, you simply type two single-quote marks
of the appropriate kind. For example, to get the phrase
\begindisplay
``I understand.''
\enddisplay
(including the quotation marks) you should type
\begintt
``I understand.''
\endtt
to your computer.

A typewriter-like style of type will be used throughout this manual to indicate
\TeX\ constructions that you might type on your terminal, so that the
symbols actually typed are readily distinguishable from the output \TeX\ would
produce and from the comments in the manual itself. Here are the symbols to be
used in the examples:
\begintt
ABCDEFGHIJKLMNOPQRSTUVWXYZ
abcdefghijklmnopqrstuvwxyz
0123456789"#$%&@*+-=,.:;?!
()<>[]{}`'\||/_^~
\endtt
If your computer terminal doesn't happen to have all of these, don't
despair; \TeX\ can make~do with the ones you have. An additional symbol
\begindisplay
\]
\enddisplay
is used to stand for a {\sl ^{blank space}}, in case it is important
to emphasize that a blank space is being typed; thus, what you {\sl
really\/} type in the example above is
\begintt
``I|]understand.''
\endtt
Without such a symbol you would have
difficulty seeing the invisible parts of certain constructions. But we
won't be using `\]' very often, because spaces are usually visible enough.

Book printing differs significantly from ordinary typing with respect to
^{dashes}, ^{hyphens}, and ^{minus signs}. In good math books,
these symbols are all different; in fact there usually are at least four
different symbols:
\begindisplay
a hyphen (-);\cr
an en-dash (--);\cr
an em-dash (---);\cr
a minus sign ($-$).\cr
\enddisplay
Hyphens are used for compound words like `daughter-in-law' and `X-rated'.
^{En-dash}es are used for number ranges like `pages 13--34', and also in
contexts like `exercise 1.2.6--52'. ^{Em-dash}es are used for punctuation in
sentences---they are what we often call simply dashes. And minus signs are
used in formulas. A conscientious user of \TeX\ will be careful to distinguish
these four usages, and here is how to do it:
\begindisplay
for a hyphen, type a hyphen (|-|);\cr
for an en-dash, type two hyphens (|--|);\cr
for an em-dash, type three hyphens (|---|);\cr
for a minus sign, type a hyphen in mathematics mode (|$-$|).\cr
\enddisplay
(Mathematics mode occurs between dollar signs; it is discussed later, so you
needn't worry about it now.)

\exercise Explain how to type the following sentence to \TeX: Alice said,
``I always use an en-dash instead of a hyphen when specifying page numbers
like `480--491' in a ^{bibliography}.''
\answer |Alice said, ``I always use an en-dash instead of a hyphen when|\break
|specifying page numbers like `480--491' in a bibliography.''| \
(The wrong answer to this question ends with |'480-49l' in a bibliography."|)

\exercise What do you think happens when you type four hyphens in a row?
\answer You get em-dash and hyphen (----), which looks awful.

If you look closely at most well-printed books, you will find that certain
combinations of letters are treated as a unit. For example, this is true
of the `f' and the `i' of `find'. Such combinations are called {\sl
^{ligatures}}, and professional typesetters have traditionally been
trained to watch for letter combinations such as |ff|, |fi|, |fl|, |ffi|, and
|ffl|. \ (The reason is that words like `f{}ind' don't look very good in
most styles of type unless a ligature is substituted for the letters that
clash. It's somewhat surprising how often the traditional ligatures appear
in English; other combinations are important in other languages.)

\exercise Think of an English word that contains two ligatures.
\answer fluffier firefly fisticuffs, flagstaff fireproofing,
chiffchaff and riffraff.

The good news is that you do
{\sl not\/} have to concern yourself with ligatures: \TeX\ is perfectly
capable of handling such things by itself, using the
same mechanism that converts `|--|' into `--'. In fact, \TeX\ will also look
for combinations of adjacent letters (like `|A|' next to `|V|'\thinspace)
that ought to be moved closer together for better appearance; this is
called {\sl ^{kerning}}.

\medbreak
To summarize this chapter: When using \TeX\ for straight copy, you type
the copy as on an ordinary typewriter, except that you need to be careful
about quotation marks, the number 1, and various kinds of hyphens/dashes.
\TeX\ will automatically take care of other niceties like ligatures and
kerning.

\danger (Are you sure you should be reading this paragraph? The
``^{dangerous bend}'' sign here is meant to warn you about material that
ought to be skipped on first reading. And maybe also on second reading.
The reader-beware paragraphs sometimes refer to concepts that aren't
explained until later chapters.)

\danger If your keyboard does not contain a left-quote symbol, you can
type ^|\lq|, followed by a space if the next character is a letter, or
followed by a |\| if the next character is a space. Similarly, ^|\rq|
yields a right-quote character. Is that clear?
\begintt
\lq\lq|]I|]understand.\rq\rq\|]
\endtt

\danger In case you need to type ^{quotes within quotes}, for example a
single quote followed by a double quote, you can't simply type
\thinspace|'''|\thinspace\ because \TeX\ will interpret this as '''
(namely, double quote followed by single quote).  If you have already read
Chapter~5, you might expect that the solution will be to use
grouping---namely, to type something like \thinspace|{'}''|. But it turns
out that this doesn't produce the desired result, because there is usually
less space following a single right quote than there is following a double
right quote: What you get is {'}'', which is indeed a single quote
followed by a double quote (if you look at it closely enough), but it
looks almost like three equally spaced single quotes.  On the other hand,
you certainly won't want to type \thinspace|'|\]|''|, because that space
is much too large---it's just as large as the space between words---and
\TeX\ might even start a new line at such a space when making up a
paragraph! The solution is to type \thinspace|'\thinspace''|, which
produces '\thinspace'' as desired.^^|\thinspace|

\dangerexercise OK, now you know how to produce ''' and '\thinspace'';
how do you get ``\thinspace` and `{}``\thinspace?
\answer |``\thinspace`|; and either |`{}``| or |{`}``| or something similar.
Reason: There's usually less space {\sl preceding\/} a single left quote than
there is preceding a double left quote. \ (Left and right are opposites.)

\dangerexercise Why do you think the author introduced the control
sequence |\thinspace| to solve the adjacent-quotes problem, instead of
recommending the trickier construction |'$|^|\,||$''| (which also works)?
\answer Eliminating ^|\thinspace| would mean that a user need not learn
the term; but it is not advisable to minimize terminology by ``overloading''
math mode with tricky constructions. For example, a user who wishes to
take advantage of \TeX's ^|\mathsurround| feature would be thwarted by
non-mathematical uses of dollar signs. \ (Incidentally, neither |\thinspace|
nor ^|\,| are built into \TeX; both are defined in terms of more
primitive features, in Appendix~B.)

\endchapter

In modern Wit all printed Trash, is
Set off with num'rous\/ {\rm Breaks}\raise.5ex\vbox{\hrule width 2em}%
  and\/ {\rm Dashes}\raise.5ex\vbox{\hrule width 1em}
% no period after the em-dash: stet!
% Sir Walter Scott ruined this quote in his edition of Swift!
\author JONATHAN ^{SWIFT}, {\sl On Poetry: A Rapsody\/} (1733) % lines 93--94
% Rapsody: stet!

\bigskip

Some compositors still object to work
in offices where type-composing machines are introduced.
\author WILLIAM STANLEY ^{JEVONS}, {\sl Political Economy\/} (1878) % sec 55
% "They are all afraid that if the work is done too easily and rapidly,
% they will not be wanted to do it."
% Jevons goes on to say that justifying and page makeup can't be done
% profitably by machines, so the employees needn't fear losing their jobs.

\eject
\beginchapter Chapter 3. Controlling\\\TeX

Your keyboard has very few keys compared to the large number of symbols that you
may want to specify. In order to make a limited keyboard sufficiently versatile,
one of the characters that you can type is reserved for special use, and
it is called the {\sl ^{escape character}}. Whenever you want to type
something that controls the format of your manuscript, or something that
doesn't use the keyboard in the ordinary way, you should type the escape
character followed by an indication of what you want to do.

Note: Some computer terminals have a key marked `|ESC|', but that is {\sl not\/}
your escape character! It is a key that sends a special message to the operating
system, so don't confuse it with what this manual calls ``escape.''

\TeX\ allows any character to be used for escapes, but the ``^{backslash}''
character `|\|' is usually adopted for this purpose, since backslashes are
reasonably convenient to type and they are rarely needed in ordinary text.
Things work out best when different \TeX\ users do things consistently,
so we shall escape via backslashes in all the examples of this manual.

Immediately after typing `|\|' (i.e., immediately after an escape
character) you type a coded command telling \TeX\ what you have in mind.
Such commands are called {\sl ^{control sequences}}. For example, you might type
^^{markup commands, see control sequences}
\begintt
\input MS
\endtt
which (as we will see later) causes \TeX\ to begin reading a file called
`|MS.tex|'; the string of characters `^|\input|' is a control sequence.
Here's another example:
\begintt
George P\'olya and Gabor Szeg\"o.
\endtt
% sic; this is the spelling used in opening pages of their famous books
% but I give the Hungarian spellings in the index
\TeX\ converts this to `George P\'olya and Gabor Szeg\"o.' There are two
^^{Polya}^^{Szego}^^{acute}^^{umlaut}
control sequences, ^|\'| and ^|\"|, here; these control sequences
have been used to place ^{accents} over some of the letters.

Control sequences come in two flavors. The first kind, like |\input|,
is called a {\sl^{control word}\/}; it
consists of an escape character followed by one or more {\sl letters}, followed
by a space or by something besides a letter. \ (\TeX\ has to know where the
control sequence ends, so you must put a space after a control word if
the next character is a letter. For example, if you type `|\inputMS|',
\TeX\ will naturally interpret this as a control word with seven
letters.) \ In case you're wondering what a ``^{letter}'' is, the answer
is that \TeX\ normally regards the 52 symbols |A...Z| and |a...z| as
letters. The digits |0...9| are {\sl not\/} considered to be
letters, so they don't appear in control sequences of the first kind.

A control sequence of the other kind, like |\'|, is called a {\sl
^{control symbol}\/}; it consists of the escape
character followed by a single {\sl nonletter}. In this case you don't
need a space to separate the control sequence from a letter that follows,
since control sequences of the second kind always have exactly one
symbol after the escape character.

\exercise What are the control sequences in `|\I'm \exercise3.1\\!|'\thinspace?
\answer |\I|, |\exercise|, and |\\|. (The last of these is of type~2, i.e.,
a control symbol, since the second backslash is not a letter; the first
backslash keeps the second one from starting its own control sequence.)

\exercise We've seen that the input |P\'olya| yields `P\'olya'. Can
you guess how the French words `math\'ematique' and `centim\`etre'
should be specified?
\answer |math\'ematique| and |centim\`etre|.^^|\'|^^|\`|

When a space comes after a control word (an all-letter control
sequence), it is ignored by
\TeX; i.e., it is not considered to be a ``real'' space belonging to the
manuscript that is being typeset. But when a space comes after a control
symbol, it's truly a space.

Now the question arises, what do you do if you actually {\sl want\/} a
space to appear after a control word? We will see later that \TeX\
treats two or more consecutive spaces as a single ^{space}, so the answer
is {\sl not\/} going to be ``type two spaces.'' The correct answer is to
type ``control space,'' ^^|\ | namely
\begintt
\|]
\endtt
(the escape character followed by a blank space); \TeX\ will treat this as
a space that is not to be ignored. Notice that |\|\] is a control
sequence of the second kind, namely a control symbol, since there is a
single nonletter (\]) following the escape character. Two consecutive
spaces are considered to be equivalent to a single space, so further
spaces immediately following |\|\] will be ignored. But if you want to
enter, say, three consecutive spaces into a manuscript you can type
`|\|\]|\|\]|\|\]'.  Incidentally, typists are often taught to put two
spaces at the ends of sentences; but we will see later that \TeX\ has its
own way to produce extra space in such cases. Thus you needn't be
consistent in the number of spaces you type.

\danger Nonprinting control characters like \<return> might follow
an escape character, and these lead to distinct control sequences according
to the rules. \TeX\ is initially set up to treat |\|\<return> and
|\|\<tab> ^^|\<return>|^^|\<tab>|
the same as |\|\] (control space); these special control sequences
should probably not be redefined, because you can't see the difference
between them when you look at them in a file.
^^{carriage-return, see <return>}

It is usually unnecessary for you to use ``control space,'' since control
sequences aren't often needed at the ends of words. But here's an example
that might shed some light on the matter: This manual itself has been
typeset by \TeX, and one of the things that occurs fairly often is the
tricky ^{logo} `\TeX', which requires backspacing and lowering the E.
There's a special control word
\begintt
\TeX
\endtt
that produces the half-dozen or so instructions necessary to typeset `\TeX'.
When a phrase like `\TeX\ ignores spaces after control words.'\ is
desired, the manuscript renders it as follows:
\begintt
\TeX\ ignores spaces after control words.
\endtt
Notice the extra |\| following ^|\TeX|; this produces the control space
that is necessary because \TeX\ ignores spaces after control words.
Without this extra |\|, the result would have been
\begindisplay
\TeX ignores spaces after control words.
\enddisplay
On the other hand, you can't simply put |\| after |\TeX| in all contexts.
For example, consider the phrase
\begintt
the logo `\TeX'.
\endtt
In this case an extra backslash doesn't work at all; in fact,
you get a curious result if you type
\begintt
the logo `\TeX\'.
\endtt
Can you guess what happens? \ Answer: The |\'| is a control sequence denoting
an acute accent, as in our |P\'olya| example above; the effect is
therefore to put an accent over the next nonblank character,
which happens to be a period. In other words, you get an accented
period, and the result is
\begindisplay
the logo `\TeX\'.
\enddisplay
Computers are good at following instructions, but not at reading your mind.

\TeX\ understands about 900 control sequences as part of its built-in
vocabulary, and all of them are explained in this manual somewhere. But
you needn't worry about learning so many different things, because you won't
really be needing very many of them unless you are faced with unusually
complicated copy. Furthermore, the ones you do need to learn actually fall into
relatively few categories, so they can be assimilated without great difficulty.
For example, many of the control sequences are simply the names of special
characters used in math formulas; you type `^|\pi|'~to get~`$\pi$',
`^|\Pi|'~to get~`$\Pi$',
`^|\aleph|'~to get~`$\aleph$',
`^|\infty|'~to get~`$\infty$',
`^|\le|'~to get~`$\le$',
`^|\ge|'~to get~`$\ge$',
`^|\ne|'~to get~`$\ne$',
`^|\oplus|'~to get~`$\oplus$',
`^|\otimes|'~to get~`$\otimes$'.
Appendix~F contains several tables of such symbols.

\danger There's no built-in relationship between ^{uppercase} and ^{lowercase}
letters in control sequence names. For example, `|\pi|' and `|\Pi|'
and `|\PI|' and `|\pI|' are four different control words.

The 900 or so control sequences that were just mentioned actually aren't
the whole story, because it's easy to define more.  For example, if you
want to substitute your own favorite names for math symbols, so that you
can remember them better, you're free to go right ahead and do it;
Chapter~20 explains how.

About 300 of \TeX's control sequences are called {\sl ^{primitive}\/}; these
are the low-level atomic operations that are not decomposable into simpler
functions. All other control sequences are defined, ultimately, in terms
of the primitive ones. For example, ^|\input| is a primitive operation,
but ^|\'| and ^|\"| are not; the latter are defined in terms of an
^|\accent| primitive.

People hardly ever use \TeX's primitive control sequences in their
manuscripts, because the primitives are $\ldots$ well $\ldots$ so
{\sl primitive}. You have to type a lot of instructions when you are
trying to make \TeX\ do low-level things; this takes time and invites
mistakes. It is generally better to make use of higher-level control
sequences that state what functions are desired, instead of typing
out the way to achieve each function each time. The higher-level control
sequences need to be defined only once in terms of primitives. For
example, |\TeX| is a control sequence that means ``typeset the \TeX\ logo'';
|\'| is a control sequence that means ``put an acute accent over the
next character''; and both of these control sequences might require different
combinations of primitives when the style of type changes. If \TeX's logo
were to change, the author would simply have to change one definition, and the
changes would appear automatically wherever they were needed. By contrast,
an enormous amount of work would be necessary to change the logo if it
were specified as a sequence of primitives each time.

At a still higher level, there are control sequences that govern the
overall format of a document. For example, in the present book the author
typed `^|\exercise|' just before stating each exercise; this |\exercise|
command was programmed to make \TeX\ do all of the following things:

\nobreak\medskip
\item\bull compute the exercise number (e.g., `3.2' for the second
exercise in Chapter~3);
\smallskip
\item\bull typeset `\thinspace{\manual\char'170\rm\kern.15em
  \ninebf EXERCISE \bf3.2}' with the appropriate typefaces, on a line by
  itself, and with the triangle sticking out in the left margin;
\smallskip
\item\bull leave a little extra space just before that line, or begin
  a new page at that line if appropriate;
\smallskip
\item\bull prohibit beginning a new page just after that line;
\smallskip
\item\bull suppress indentation on the following line.
\medbreak\noindent
It is obviously advantageous to avoid typing all of these individual
instructions each time. And since the manual is entirely described in
terms of high-level control sequences, it could be printed in a radically
different format simply by changing a dozen or so definitions.
% and sweating over the page layout in the math and alignment chapters!

\danger How can a person distinguish a \TeX\ primitive from a control sequence
that has been defined at a higher level? There are two ways: \ (1)~The index
to this manual lists all of the control sequences that are discussed, and each
primitive is marked with an asterisk. \ (2)~You can display the meaning of a
control sequence while running \TeX\null. If you type `^|\show||\cs|'
where |\cs| is any control sequence, \TeX\ will respond with its current
meaning. For example, `|\show\input|' results in \hbox{`|> \input=\input.|'},
because |\input| is primitive. On the other hand, `|\show|^|\thinspace|' yields
\begintt
> \thinspace=macro:
->\kern .16667em .
\endtt
This means that |\thinspace| has been defined as an abbreviation for
`|\kern|~|.16667em|~'. By typing `|\show|\penalty0|\kern|' you can verify
that ^|\kern| is primitive. The results of\/ |\show| appear on your
terminal and in the ^{log file} that you get after running \TeX.

\dangerexercise Which of the control sequences |\|\] and
|\|\<return> is primitive?
\answer According to the index, |\|\] is primitive but
|\|\<return> isn't. The command `|\def\^^M{\ }|' in
Appendix~B is what actually defines |\|\<return>, since a
return is representable as |^^M|. Asking \TeX\ to |\show\^^M|
\looseness-1
produces the response `|>| |\^^M=macro:->\|\]|.|'.

In the following chapters we shall frequently discuss ``^{plain \TeX}''
format, which is a set of about 600 ^{basic control sequences} that are
defined in Appendix~B\null. These control sequences, together with the 300
or so primitives, are usually present when \TeX\ begins to process a
manuscript; that is why \TeX\ claims to know roughly 900 control sequences
when it starts.  We shall see how plain \TeX\ can be used to create
documents in a flexible format that meets many people's needs, using some
typefaces that come with the \TeX\ system. However, you should keep in
mind that plain \TeX\ is only one of countless ^{formats} that can be
designed on top of \TeX's primitives; if you want some other format, it
will usually be possible to adapt \TeX\ so that it will handle whatever
you have in mind. The best way to learn is probably to start with plain
\TeX\ and to change its definitions, little by little, as you gain more
experience.

\danger Appendix E contains examples of formats that can be added to
Appendix~B for special applications; for example, there is a set of
definitions suitable for business correspondence. A complete specification
of the format used to typeset this manual also appears in Appendix~E\null.
Thus, if your goal is to learn how to design \TeX\ formats, you will
probably want to study Appendix~E while mastering Appendix~B\null. After you
have become skilled in the lore of control-sequence definition, you
will probably have developed some formats that other people will want
to use; you should then write a supplement to this manual, explaining
your style rules.

The main point of these remarks, as far as novice \TeX\ users are concerned, is
that it is indeed possible to define nonstandard \TeX\ control sequences.
When this manual says that something is part of ``plain \TeX,'' it means
that \TeX\ doesn't insist on doing things exactly that way; a person
could change the rules by changing one or more of the definitions in
Appendix~B\null. But you can safely rely on the control sequences of plain
\TeX\ until you become an experienced \TeX nical~typist.

\ddangerexercise How many different control sequences of length~2
(including the escape character) are possible? How many of length~3?
\answer There are 256 of length~2; most of these are undefined when \TeX\
begins. \ (\TeX\ allows any character to be an escape, but it does not
distinguish between control sequences that start with different escape
characters.) \
If we assume that there are 52 letters, there are exactly $52^2$
possible control sequences of length~3 (one for each pair of letters, from
|AA| to |zz|). But Chapter~7 explains how to use ^|\catcode| to change any
character into a ``^{letter}''; therefore it's possible to use any of
$256^2$ potential control sequences of length~3.

\endchapter

Syllables govern the world.
\author JOHN ^{SELDEN}, {\sl Table Talk\/} (1689) % section on Power

\bigskip

I claim not to have controlled events,
but confess plainly that events have controlled me.
\author ABRAHAM ^{LINCOLN} (1864) % letter to A. G. Hodges, April 4

\eject
\beginchapter Chapter 4. Fonts\\of Type

Occasionally you will want to change from one ^{typeface} to another, for
example if you wish to be {\bf ^{bold}} or to {\sl emphasize\/} something.
\TeX\ deals with sets of up to 256 characters called ``^{fonts}'' of type,
and control sequences are used to select a particular font. For example,
you could specify the last few words of the first sentence above
in the following way, using the plain \TeX\ format of Appendix~B:
\begintt
to be \bf bold \rm or to \sl emphasize \rm something.
\endtt
Plain \TeX\ provides the following control sequences for changing fonts:
\begindisplay
^|\rm| switches to the normal ``roman'' typeface:&Roman\cr
^|\sl| switches to a slanted roman typeface:&\sl Slanted\cr
^|\it| switches to italic style:&\it Italic\cr
^|\tt| switches to a typewriter-like face:&\tt Typewriter\cr
^|\bf| switches to an extended boldface style:&\bf Bold\cr
\enddisplay
^^{typewriter type}^^{face}
At the beginning of a run you get ^{roman type} (|\rm|) unless you specify
otherwise.

Notice that two of these faces have an ``^{oblique}'' slope for emphasis:
{\sl ^{Slanted type} is essentially the same as roman, but the letters are
slightly skewed, \it while the letters in ^{italic type} are drawn in a
different style.} \ (You can perhaps best appreciate the difference between
the roman and italic styles by contemplating {\tenu letters that are
in an unslanted italic face.}) \ Typographic conventions are presently
in a state of transition, because new technology has made it possible to
do things that used to be prohibitively expensive; people are wrestling
with the question of how much to use their new-found typographic freedom.
Slanted roman type was introduced in the 1930s, but it first became widely
used as an alternative to the conventional italic during the late 1970s.
It can be bene\-ficial in mathematical texts, since slanted
letters are distinguishable from the italic letters in math formulas.
The double use of italic type for two different purposes---for example,
when statements of theorems are italicized as well as the names of variables in
those theorems---has led to some confusion, which can now be
avoided with slanted type. People are not generally agreed about the relative
merits of slanted versus italic, but slanted type is rapidly becoming a
favorite for the titles of books and journals in bibliographies.

Special fonts are effective for emphasis, but not for sustained reading;
your eyes would tire if long portions of this manual were entirely set in
a bold or slanted or italic face. Therefore roman type accounts for the
bulk of most typeset material. But it's a nuisance to say `|\rm|' every
time you want to go back to the roman style, so \TeX\ provides an easier
way to do it, using ``^{curly brace}^^{brace}'' symbols: You can switch
fonts inside the special symbols |{| and |}|, without affecting the fonts
outside. For example, the displayed phrase at the beginning of this
chapter is usually rendered
\begintt
to be {\bf bold} or to {\sl emphasize} something.
\endtt
This is a special case of the general idea of ``^{grouping}'' that we shall
discuss in the next chapter. It's best to forget about the first way of
changing fonts, and to use grouping instead; then your \TeX\ manuscripts
will look more natural, and you'll probably
never\footnote*{Well \dots, hardly ever.} have to type `|\rm|'.

\exercise Explain how to type the bibliographic reference `Ulrich ^{Dieter},
{\sl Journal f\"ur die reine und angewandte Mathematik\/ \bf201} (1959),
37--70.' [Use grouping.]
\answer |Ulrich Dieter, {\sl Journal f\"ur die reine und angewandte|\parbreak
        |Mathematik\/ \bf201} (1959), 37--70.|\par\nobreak\smallskip\noindent
It's convenient to use a single group for both |\sl| and |\bf| here. The
`|\/|' is a refinement that you might not understand until you read the
rest of Chapter~4.

We have glossed over an important aspect of quality in the preceding
discussion. Look, for example, at the {\it italicized} and {\sl slanted}
words in this sentence. Since italic and slanted styles slope to the right,
the d's stick into the spaces that separate these words from the roman
type that follows; as a result, the spaces appear to be too skimpy,
although they are correct at the base of the letters. To equalize the
effective white space, \TeX\ allows you to put the special control sequence
`^|\/|' just before switching back to unslanted letters. When you type
\begintt
{\it italicized\/} and {\sl slanted\/} words
\endtt
you get {\it italicized\/} and {\sl slanted\/} words that look better.
The `|\/|' tells \TeX\ to add an\break % makes the line tighter, to be fair
``{\sl^{italic correction}\/}'' to the
previous letter, depending on that letter; this correction is about four
times as much for an `$f$' as for a `$c$', in a typical italic font.

Sometimes the italic correction is not desirable, because other factors take
up the visual slack. The standard rule of thumb is to use |\/| just before
switching from slanted or italic to roman or bold, unless the next
character is a period or comma. For example, type
\begintt
{\it italics\/} for {\it emphasis}.
\endtt
Old manuals of style say that the ^{punctuation} after a word should be in the
{\it same\/} font as that {\it word;\/} but an italic semicolon often looks
wrong, so this convention is changing. When an italicized word occurs
just before a semicolon, the author recommends typing `|{\it word\/};|'.

\exercise {\it Explain how to typeset a\/ {\rm roman} word in the midst
of an italicized sentence.}
\answer |{\it Explain ... typeset a\/ {\rm roman} word ... sentence.}|
Note the position of the italic correction in this case.

\danger Every letter of every font has an italic correction, which you can
bring to life by typing |\/|. The correction is usually zero in unslanted
styles, but there are exceptions: To typeset a bold `{\bf f\/}' in quotes,
you should say |a| |bold| \hbox{|`{\bf f\/}'|}, lest you get a bold `{\bf f}'.

\ddangerexercise Define a control sequence |\ic| such that `|\ic c|' puts the
italic correction of character $c$ into \TeX's register |\dimen0|.
\answer |\def\ic#1{\setbox0=\hbox{#1\/}\dimen0=\wd0|\parbreak
|\setbox0=\hbox{#1}\advance\dimen0 by -\wd0 }|.

\ddanger The primitive control sequence ^|\nullfont| stands for a font that
has no characters. This font is always present, in case you haven't
specified any others.

Fonts vary in size as well as in shape. For example, the font you are now
reading is called a ``10-point'' font, because certain features of its
design are 10 ^{points} apart, when measured in printers' units. \ (We
will study the point system later; for now, it should suffice to point out
that the parentheses around this sentence are exactly 10 points tall---and
the em-dash is just 10 points wide.) \ The ``^{dangerous bend}''
sections of this manual are set in 9-point type, the foot\-notes in 8-point,
^{subscripts} in 7-point or 6-point, sub-subscripts in 5-point.

Each font used in a \TeX\ manuscript is associated with a control sequence;
for example, the 10-point font in this paragraph is called ^|\tenrm|, and
the corresponding 9-point font is called ^|\ninerm|. The slanted fonts that
match |\tenrm| and |\ninerm| are called ^|\tensl| and ^|\ninesl|. These
control sequences are not built into \TeX, nor are they the actual names
of the fonts; \TeX\ users are just supposed to make up convenient names,
whenever new fonts are introduced into a manuscript. Such control
sequences are used to change typefaces.

When fonts of different sizes are used simultaneously, \TeX\ will line the
letters up according to their ``^{baseline}s.'' For example, if you type
\begintt
\tenrm smaller \ninerm and smaller
\eightrm and smaller \sevenrm and smaller
\sixrm and smaller \fiverm and smaller \tenrm
\endtt
the result is {smaller \ninerm and smaller \eightrm and smaller
\sevenrm and smaller \sixrm and smaller \fiverm and smaller}. Of course
this is something that authors and readers aren't accustomed to, because
printers couldn't do such things with traditional lead types. Perhaps
poets who wish to speak in {\fiverm a still small voice} will cause future
books to make use of frequent font variations, but nowadays it's only
an occasional font freak {\fiverm(like the author of this manual)} who
likes such experiments. One should not get too carried away by the prospect
of font switching unless there is good reason.

An alert reader might well be confused at this point because we started out
this chapter by saying that `|\rm|' is the command that switches to roman
type, but later on we said that `|\tenrm|' is the way to do it. The truth
is that both ways work. But it has become customary to set things up so that
|\rm| means ``switch to roman type in the current size'' while |\tenrm| means
``switch to roman type in the 10-point size.'' In plain \TeX\ format, nothing
but 10-point fonts are provided, so |\rm| will always get you |\tenrm|; but
in more complicated formats the meaning of\/ |\rm| will change in different
parts of the manuscript. For example, in the format used by the author to
typeset this manual, there's a control sequence `^|\tenpoint|' that causes
|\rm| to mean |\tenrm|, |\sl| to mean |\tensl|, and so on, while
`^|\ninepoint|' changes the definitions so that |\rm| means |\ninerm|,
etc. There's another control sequence used to introduce the quotations at
the end of each chapter; when the quotations are typed, |\rm| and |\sl|
temporarily stand for {\eightss 8-point unslanted sans-serif type} and
{\eightssi 8-point slanted sans-serif type}, respectively. This device of
constantly redefining the abbreviations |\rm| and |\sl|, behind the
scenes, frees the typist from the need to remember what size or style of
type is currently being used.

\exercise Why do you think the author chose the names `|\tenpoint|' and
`|\tenrm|', etc., instead of `|\10point|' and `|\10rm|'\thinspace?
\answer Control word names are made of letters, not digits.

\dangerexercise Suppose that you have typed a manuscript using slanted type for
emphasis, but your editor suddenly tells you to change all the slanted to
italic.  What's an easy way to do this?
\answer Say |\def\sl{\it}| at the beginning, and delete other definitions
of\/ |\sl| that might be present in your format file (e.g., there might be
one inside a |\tenpoint| macro).

\danger Each font has an external name that identifies it with respect to
all other fonts in a particular library. For example, the font in this
sentence is called `|cmr9|', which is an abbreviation for ``^{Computer
Modern} Roman 9~point.'' ^^{cm fonts} In order to prepare \TeX\ for
using this font, the command
\begintt
\font\ninerm=cmr9
\endtt
appears in Appendix E\null. In general you say `^|\font||\cs=|\<external
font name>' to load the information about a particular font into \TeX's
memory; afterwards the control sequence |\cs| will select that font for
typesetting.  Plain \TeX\ makes only sixteen fonts available initially (see
Appendix~B and Appendix~F\null), but you can use |\font| to access
anything that exists in your system's font library.

\danger It is often possible to use a font at several different sizes, by
magnifying or shrinking the character images. Each font has a so-called
^{design size}, which reflects the size it normally has by default; for
example, the design size of |cmr9| is 9~points. But on many systems there is
also a range of sizes at which you can use a particular font, by scaling its
dimensions up or down. To load a scaled font into \TeX's memory, you
simply say `|\font\cs=|\<external font name> ^|at| \<desired size>'.
For example, the command
\begintt
\font\magnifiedfiverm=cmr5 at 10pt
\endtt
brings in 5-point Computer Modern Roman at twice its normal size. \ (Caution:
Before using this `|at|' feature, you should check to make sure that your
typesetter supports the font at the size in question; \TeX\ will accept any
\<desired size> that is positive and less than 2048 points, but the final
output will not be right unless the scaled font really is available on your
printing device.)

\danger What's the difference between |cmr5| |at| |10pt| and the normal
10-point font, |cmr10|? Plenty; a well-designed font will be drawn
differently at different point sizes, and the letters will often have
different relative heights and widths, in order to enhance readability.
\begindisplay
\tenrm Ten-point type is different from%
  \magnifiedfiverm\ magnif{}ied f{}ive-point type.
\enddisplay
It is usually best to scale fonts only slightly with respect to
their design size, unless the final product is going to be photographically
reduced after \TeX\ has finished with it, or unless you are trying for an
unusual effect.^^{magnification}^^{reduction}

\danger Another way to magnify a font is to specify a scale factor that is
relative to the design size. For example, the command
\begintt
\font\magnifiedfiverm=cmr5 scaled 2000
\endtt
is another way to bring in the font ^|cmr5| at double size. The scale factor
is specified as an integer that represents a magnification ratio times~1000.
Thus, a scale factor of 1200 specifies magnification by 1.2, etc.

\dangerexercise State two ways to load font |cmr10| into \TeX's memory
at half its normal size.
\answer |\font\squinttenrm=cmr10 at 5pt|\parbreak
        |\font\squinttenrm=cmr10 scaled 500|

\font\onerm=cmr10 scaled\magstep1
\font\onett=cmtt10 scaled\magstep1
\font\tworm=cmr10 scaled\magstep2
\font\twott=cmtt10 scaled\magstep2
%\font\threerm=cmr10 scaled\magstep3 % such large magnifications look ugly
%\font\threett=cmtt10 scaled\magstep3 % in a book context!
\danger At many computer centers it has proved convenient to supply fonts
at magnifications that grow in geometric ratios---something like equal-tempered
tuning on a ^{piano}. The idea is to have all fonts available at their true
size as well as at magnifications 1.2 and~1.44 (which is $1.2\times1.2$);
perhaps also at magnification~1.728 ($=1.2\times1.2\times1.2$) and even
higher. Then you can magnify an entire document by 1.2 or~1.44 and still
stay within the set of available fonts. Plain \TeX\ provides the
abbreviations ^|\magstep||0| for a scale factor of 1000, |\magstep1| for a
scaled factor of 1200, |\magstep2| for 1440, and so on up to |\magstep5|.
You say, for example,
\begintt
\font\bigtenrm=cmr10 scaled\magstep2
\endtt
to load font |cmr10| at $1.2\times1.2$ times its normal size.
\begindisplay \lineskip5pt
\tenrm\llap{``}This is {\tentt cmr10} at normal size
  ({\tentt \char`\\magstep0}).''\cr
\onerm\llap{``}This is {\onett cmr10} scaled once by 1.2
  ({\onett \char`\\magstep1}).''\cr
\tworm\llap{``}This is {\twott cmr10} scaled twice by 1.2
  ({\twott \char`\\magstep2}).''\cr
%\threerm\llap{``}This is {\threett cmr10} scaled by
%  {\threett \char`\\magstep3}.''\cr
\enddisplay
(Notice that a little magnification goes a long way.) \
There's also ^|\magstephalf|, which magnifies by $\sqrt{1.2}$,
i.e., halfway between steps 0 and~1.

\danger Chapter~10 explains how to apply magnification to an entire
document, over and above any magnification that has been specified when
fonts are loaded. For example, if you have loaded a font that is scaled
by |\magstep1| and if you also specify ^|\magnification||=\magstep2|, the
actual font used for printing will be scaled by |\magstep3|. Similarly, if
you load a font scaled by |\magstephalf| and if you also say
|\magnification=\magstephalf|, the printed results will be
scaled by |\magstep1|.

\endchapter

Type faces---like people's faces---have distinctive features
indicating aspects of character. % I don't think he was kidding
\author MARSHALL ^{LEE}, {\sl Bookmaking\/} (1965) % page 83

\bigskip

This was the Noblest Roman of them all.
\author WILLIAM ^{SHAKESPEARE}, {\sl The Tragedie %
  of Julius C\ae sar\/} (1599) % Act V, Scene 5, line 68
  % For Shakespeare I'm using the spelling from First Folio (1623)
  % (titles not from the contents page, but the running heads in the plays)
  % but act/line numbers from The Riverside Shakespeare (throughout)

\eject
\beginchapter Chapter 5. Grouping

Every once in a while it is necessary to treat part of a manuscript as a
unit, so you need to indicate somehow where that part begins and where it
ends. For this purpose \TeX\ gives special interpretation to two
``^{grouping characters},'' which (like the escape character) are
treated differently from the normal symbols that you type. We assume in
this manual that |{| and |}| are the grouping characters, since they
are the ones used in plain \TeX. ^^{curly braces, see braces}

We saw examples of grouping in the previous chapter, where it was mentioned
that font changes inside a group do not affect the fonts in force outside.
The same principle applies to almost anything else that is defined inside
a group, as we will see later; for example, if you define a control sequence
within some group, that definition will disappear when the group ends.
In this way you can conveniently instruct \TeX\ to do something unusual,
by changing its normal conventions temporarily inside of a group; since
the changes are invisible from outside the group, there is no need to worry
about messing up the rest of a manuscript by forgetting to restore the
normal conventions when the unusual construction has been finished.
Computer scientists have a name for this aspect of grouping, because it's
an important aspect of programming languages in general; they call it
``^{block structure},'' and definitions that are in force only within
a group are said to be ``^{local}'' to that group.

You might want to use grouping even when you don't care about block
structure, just to have better control over spacing. For example, let's
consider once more the control sequence ^|\TeX| that produces
the logo `\TeX' in this manual: We observed in Chapter~3 that a blank space
after this control sequence will be gobbled up unless one types
`\hbox{|\TeX\ |}', yet it is a mistake to say `|\TeX\|' when the following
character is not a blank space. Well, in {\sl all\/} cases it would be
correct to specify the simple group
\begintt
{\TeX}
\endtt
whether or not the following character is a ^{space}, because the |}| stops
\TeX\ from absorbing an optional space into |\TeX|. This might come in
handy when you're using a text editor (e.g., when replacing all occurrences
of a particular word by a control sequence). Another thing you could do is
type
\begintt
\TeX{}
\endtt
using an {\sl empty\/} group for the same purpose: The `|{}|' here is a
group of no characters, so it produces no output, but it does have the
effect of stopping \TeX\ from skipping blanks.
^^{empty group} ^^{lbrace rbrace}

\exercise Sometimes you run into a rare word like `shelfful' that looks
better as `shelf{}ful' without the `ff' ^{ligature}. How can you fool
\TeX\ into thinking that there aren't two consecutive f's in such a word?
\answer |{shelf}ful| or |shelf{}ful|, etc.; or even |shelf\/ful|, which
yields a shelf\/ful instead of a shelf{\kern0pt}ful.
In fact, the latter idea---to
insert an ^{italic correction}---is preferable because \TeX\ will ^^|\/|
reinsert the ff ligature by itself after ^{hyphenating} |shelf{}ful|. \
(Appendix~H points out that ligatures are put into a hyphenated word that
contains no ``^{explicit kerns},'' and an italic correction is an
explicit kern.) \ But the italic correction may be too much (especially in an
italic font); |shelf{|^|\kern||0pt}ful| is often best.

\dangerexercise Explain how to get three blank spaces in a row without
using `|\|\]'.^^{control space}
\answer `\]|{|\]|}|\]' or `\]|{}|\]|{}|\]', etc. Plain \TeX\ also has a
^|\space| macro, so you can type |\space\space\space|.  \ (These aren't
strictly equivalent to `|\|\]|\|\]|\|\]', since they adjust the spaces by
the current ``^{space factor},'' as explained later.)

But \TeX\ also uses grouping for another, quite different, purpose, namely
to determine how much of your text is to be governed by certain control
sequences. For example, if you want to center something on a line you can type
\begintt
\centerline{This information should be centered.}
\endtt
using the control sequence ^|\centerline| defined in plain \TeX\ format.

Grouping is used in quite a few of \TeX's more intricate instructions; and
it's possible to have groups within groups within groups, as you can see
by glancing at Appendix~B\null. Complex grouping is generally unnecessary,
however, in ordinary manuscripts, so you needn't worry about it. Just
don't forget to finish each group that you've started, because a lost
`|}|' might cause trouble.

Here's an example of two groups, one ^{nested} inside the other:
\begintt
\centerline{This information should be {\it centered}.}
\endtt
As you might expect, \TeX\ will produce a centered line that also contains
italics:
$$\hbox{This information should be {\it centered}.}$$
But let's look at the example more closely: `|\centerline|' appears outside
the curly braces, while `|\it|' appears inside. Why are the two cases
different? And how can a beginner learn to remember which is which?
Answer: |\centerline| is a control sequence that applies only to the very next
thing that follows, so you want to put braces around the text that is to
be centered (unless that text consists of a single symbol or control sequence).
For example, to center the \TeX\ logo on a line, it would suffice to
type `|\centerline\TeX|', but to center the phrase `\TeX\ has groups' you
need braces: `|\centerline{\TeX\ has groups}|'. On the other hand, |\it| is
a control sequence that simply means ``change the current font''; it acts
without looking ahead, so it affects {\sl everything\/} that follows, at
least potentially. The braces surround |\it| in order to confine the font
change to a local region.

In other words, the two sets of braces in this example actually have different
functions: One serves to treat several words of the text as if they
were a single object, while the other provides local block structure.

\exercise What do you think happens if you type the following:
\begintt
\centerline{This information should be {centered}.}
\centerline So should this.
\endtt
\answer In the first case, you get the same result as if the innermost
braces had not appeared at all, because you haven't used the grouping to
change fonts or to control spacing or anything. \TeX\ doesn't mind if you
want to waste your time making groups for no particular reason.
But in the second case, the necessary braces were forgotten. You get the
letter `S' centered on a line by itself, followed by a paragraph that
begins with `o should this.'\ on the next line.

\exercise And how about this one?
\begintt
\centerline{This information should be \it centered.}
\endtt
\answer You get the same result as if another pair of braces were present
around `|\it centered|', except that the period is typeset from the
italic font. \ (Both periods look about the same.) \ The |\it| font
will not remain in force after the |\centerline|, but this is
something of a coincidence: \TeX\ uses the braces to determine what
text is to be centered, but then it removes the braces. The
|\centerline| operation, as defined in Appendix~B\null, puts the
resulting braceless text inside {\sl another\/} group; and that's why
|\it| disappears after |\centerline|. \ (If you don't understand this,
just don't risk leaving out braces in tricky situations, and you'll be OK.)

\smallskip
\dangerexercise Define a control sequence |\ital| so that a user could type
`|\ital{text}|' instead of `|{\it text\/}|'. Discuss the pros and cons of
|\ital| versus |\it|.
\answer |\def\ital#1{{\it#1\/}}|. \ Pro:~Users might find this easier to
learn, because it works more like |\centerline| and they don't have to
remember to make the italic correction. \ Con:~To avoid the italic correction
just before a {\it comma} or {\it period}, users should probably be taught
another control sequence; for example, with
\begintt
\def\nocorr{\kern0pt }
\endtt
a user could type `|\ital{comma} or \ital{period\nocorr},|'. The alternative
of putting a period or comma in italics, to avoid the italic correction,
doesn't look as good. A long sequence of italics would be inefficient for
\TeX, since the entire text for the argument to |\ital| must be read into
memory only to be scanned again.

\ddanger Subsequent chapters describe many primitive operations of \TeX\ for
which the locality of grouping is important. For example, if one of \TeX's
internal parameters is changed within a group, the previous contents of
that parameter will be restored when the group ends. Sometimes, however,
it's desirable to make a definition that transcends its current group. This
effect can be obtained by prefixing `^|\global|' to the definition. For
example, \TeX\ keeps the current page number in a register called~|\count0|,
and the routine that outputs a page wants to increase the
^{page number}. ^{Output routines} are always protected by enclosing them
in groups, so that they do not inadvertently mess up the rest of \TeX; but
the change to |\count0| would disappear if it were kept local to the
output group. The command
\begintt
\global\advance\count0 by 1
\endtt
^^|\advance|
solves the problem; it increases |\count0| and makes this value stick around
at the end of the output routine. In general, |\global| makes the immediately
following definition pertain to all existing groups, not just to the
innermost one.

\ddangerexercise If you think you understand local and global definitions,
here's a little test to make sure: Suppose |\c| stands for `|\count1=|',
|\g| stands for `|\global\count1=|', and |\s| stands for
`|\showthe\count1|'. What values will be shown?
\begintt
{\c1\s\g2{\s\c3\s\g4\s\c5\s}\s\c6\s}\s
\endtt
\answer |{1 {2 3 4 5} 4 6} 4|.

\ddanger Another way to obtain block structure with \TeX\ is to use the
primitives ^|\begingroup| and ^|\endgroup|. These control sequences make it
easy to begin a group within one control sequence and end it within
another. The text that \TeX\ actually executes, after control sequences
have been expanded, must have properly ^{nested groups}, i.e., groups that
don't overlap. For example,
\begintt
{ \begingroup } \endgroup
\endtt
is not legitimate.

\ddangerexercise Define control sequences |\beginthe|\<block name> and
|\endthe|\<block name> that provide a ``named'' block structure. In other
words,
\begintt
\beginthe{beguine}\beginthe{waltz}\endthe{waltz}\endthe{beguine}
\endtt
should be permissible, but not
\begintt
\beginthe{beguine}\beginthe{waltz}\endthe{beguine}\endthe{waltz}.
\endtt
\answer |\def\beginthe#1{\begingroup\def\blockname{#1}}|\parbreak
|\def\endthe#1{\def\test{#1}%|\parbreak
|  \ifx\test\blockname\endgroup|\parbreak
|  \else\errmessage{You should have said|\parbreak
|    \string\endthe{\blockname}}\fi}|

\endchapter

I have had recourse to varieties of type,
and to braces.
\author JAMES ^{MUIRHEAD}, {\sl The Institutes of Gaius\/} (1880) % p. xii
% he actually said braces for what we call brackets

\bigskip

An encounter group is a gathering, for a few hours or a few days,
of twelve or eighteen personable, responsible, certifiably normal
and temporarily smelly people.
\author JANE ^{HOWARD}, {\sl Please Touch\/} (1970)

\eject
\beginchapter Chapter 6. Running\\\TeX

The best way to learn how to use \TeX\ is to use it. Thus, it's high time
for you to sit down at a computer terminal and interact with the \TeX\
system, trying things out to see what happens. Here are some small but
complete examples suggested for your first encounter.
^^{Running the program}

Caution: This chapter is rather a long one. Why don't you stop reading
now, and come back fresh tomorrow?

\smallskip OK, let's suppose that you're rested and excited about having a
trial run of \TeX\null. Step-by-step instructions for using it appear in this
chapter. First do this: Go to the lab where the graphic output device is,
since you will be wanting to see the output that you get---it won't really
be satisfactory to run \TeX\ from a remote location, where you can't hold
the generated documents in your own hands. Then log in; and start \TeX. \
(You may have to ask somebody how to do this on your local computer. Usually
the operating system prompts you for a command and you type `|tex|' or
`|run| |tex|' or something like that.)

When you're successful, \TeX\ will welcome you with a message such as
\begintt
This is TeX, Version 3.141 (preloaded format=plain 89.7.15)
**
\endtt
The `^|**|' is \TeX's way of asking you for an input file name.
% Incidentally, 89.7.15 was Jill's 50th birthday.

Now type `^|\relax|' (including the backslash), and ^\<return>
(or whatever is used to mean ``end-of-line'' on your terminal).
\TeX\ is all geared up for action, ready to read a long manuscript; but
you're saying that it's all right to take things easy, since this is
going to be a real simple run. In fact, |\relax| is a control sequence
that means ``do nothing.''

The machine will type another asterisk at you. This time type something
like `|Hello?|'\ and wait for another ^{asterisk}. Finally type `^|\end|',
and stand back to see what happens.

\TeX\ should respond with `^|[1]|' (meaning that it has finished page~1
of your output); then the program will halt, probably with some
indication that it has created a file called `|texput.dvi|'. \ (\TeX\
uses the name ^|texput| for its output when you haven't specified any
better name in your first line of input; and ^|dvi| stands for
``^{device independent},'' since |texput.dvi| is capable of
being printed on almost any kind of typographic output device.)

Now you're going to need some help again from your friendly local
computer hackers. They will tell you how to produce hardcopy from
|texput.dvi|. And when you see the hardcopy---Oh, glorious day!---you
will see a magnificent `Hello?'\ and the page number `1' at the bottom.
Congratulations on your first masterpiece of fine printing.

\smallbreak
The point is, you understand now how to get something through the whole cycle.
It only remains to do the same thing with a somewhat longer document.
So our next experiment will be to work from a file instead of typing
the input online.

Use your favorite text editor to create a file called ^|story.tex|
that contains the following 18 lines of text (no more, no less):
$$\halign{\hbox to\parindent{\hfil\sevenrm#\ \ }&#\hfil\cr
1&|\hrule|\cr\noalign{^^|\hrule|}
2&|\vskip 1in|\cr\noalign{^^|\vskip|^^{leading, see vskip}}
3&|\centerline{\bf A SHORT STORY}|\cr\noalign{^^|\centerline|}
4&|\vskip 6pt|\cr
5&|\centerline{\sl by A. U. Thor}|\cr\noalign{^^{Thor}}
6&|\vskip .5cm|\cr
7&|Once upon a time, in a distant|\cr
8&|  galaxy called \"O\"o\c c,|\cr\noalign{^^|\"|^^|\c|}
9&|there lived a computer|\cr
10&|named R.~J. Drofnats.|\cr\noalign{^^{Drofnats}}
11&||\cr
12&|Mr.~Drofnats---or ``R. J.,'' as|\cr
13&|he preferred to be called---|\cr
14&|was happiest when he was at work|\cr
15&|typesetting beautiful documents.|\cr
16&|\vskip 1in|\cr
17&|\hrule|\cr
18&|\vfill\eject|\cr\noalign{^^|\vfill|^^|\eject|}}$$
\write16{\ifnum\pageno=\storypage
  \else Redefine \string\storypage to be \the\pageno\fi}%
(Don't type the numbers at the left of these lines, of course; they are present
only for reference.) \ This example is a bit long, and more than a bit silly;
but it's no trick for a good typist like you and it will give you some
worthwhile experience, so do it. For your own good. And think about what
you're typing, as you go; the example introduces a few important features
of \TeX\ that you can learn as you're making the file.

Here is a brief explanation of what you have just typed: Lines 1 and~17
put a horizontal ^{rule} (a thin line) across the page. Lines 2 and~16
skip past one inch of space; `|\vskip|' means ``vertical skip,'' and this
extra space will separate the horizontal rules from the rest of the copy.
Lines 3 and~5 produce the title and the author name, centered, in boldface
and in slanted type. Lines 4 and~6 put extra white space between those
lines and their successors. \ (We shall discuss units of measure like
`|6pt|' and `|.5cm|' in Chapter~10.)

The main bulk of the story appears on lines 7--15, and it consists of
two ^{paragraphs}. The fact that line~11 is blank informs \TeX\ that
^^{blank line} ^^{empty line}
line~10 is the end of the first paragraph; and the `|\vskip|' on line~16
implies that the second paragraph ends on line~15, because vertical
skips don't appear in paragraphs. Incidentally, this example seems
to be quite full of \TeX\ commands; but it is atypical in that respect,
because it is so short and because it is supposed to be teaching things.
Messy constructions like |\vskip| and |\centerline| can be expected at the
very beginning of a manuscript, unless you're using a canned format, but
they don't last long; most of the time you will find yourself typing
straight text, with relatively few control sequences.

And now comes the good news, if you haven't used computer typesetting
before:  You don't have to worry about where to break lines in a paragraph
(i.e., where to stop at the right margin and to begin a new line), because
\TeX\ will do that for you. Your manuscript file can contain long lines or
short lines, or both; it doesn't matter. This is especially helpful when
you make changes, since you don't have to retype anything except the words
that changed. {\sl Every time you begin a new line in your manuscript file
it is essentially the same as typing a space.} When \TeX\ has read an
entire paragraph---in this case lines 7 to~11---it will try to break up
the text so that each line of output, except the last, contains about the
same amount of copy; and it will hyphenate words if necessary to keep the
spacing consistent, but only as a last resort.

Line 8 contains the strange concoction
\begintt
\"O\"o\c c
\endtt
and you already know that |\"| stands for an ^{umlaut} accent. The
|\c| stands for a ``^{cedilla},'' so you will get `\"O\"o\c c' as the
name of that distant galaxy.

The remaining text is simply a review of the conventions that we discussed
long ago for dashes and quotation marks, except that the `|~|' signs in
lines 10 and~12 are a new wrinkle. These are called {\sl ^{ties}}, because
they tie words together; i.e., \TeX\ is supposed to treat `|~|' as a
normal space but not to break between lines there.
A good typist will use ties within names, as shown in our
example; further discussion of ties appears in Chapter~14. ^^{tilde}

Finally, line~18 tells \TeX\ to `^|\vfill|', i.e., to fill the rest of
the page with white space; and to `^|\eject|' the page, i.e., to send it
to the output file.

\smallskip Now you're ready for Experiment~2: Get \TeX\ going again.
This time when the machine says `|**|' you should answer `|story|', since
that is the name of the file where your input resides. \ (The file
could also be called by its full name `|story.tex|', but \TeX\ automatically
supplies the suffix `|.tex|' if no suffix has been specified.)
^^{file names}

You might wonder why the first prompt was `^|**|', while the subsequent
ones are `^|*|'; the reason is simply that the first thing you type to
\TeX\ is slightly different from the rest: If the first character of your
response to `|**|' is not a backslash, \TeX\ automatically inserts
`^|\input|'. Thus you can usually run \TeX\ by merely naming your input
file. \ (Previous \TeX\ systems required you to start by typing `|\input
story|' instead of `|story|', and you can still do that; but most \TeX\
users prefer to put all of their commands into a file instead of typing
them online, so \TeX\ now spares them the nuisance of starting out with
|\input| each time.) \ Recall that in Experiment~1 you typed `|\relax|';
that started with a backslash, so |\input| was not implied.

\danger There's actually another difference between `|**|' and `|*|': If the
first character after |**| is an ^{ampersand} (\thinspace`|&|'\thinspace),
\TeX\ will replace its memory with a precomputed ^{format file} before
proceeding. Thus, for example, you can type `|&plain \input story|' or
even `|&plain story|' in response to `|**|', if you are running some
version of \TeX\ that might not have the plain format preloaded.
^^{preloaded formats}

\danger Incidentally, many systems allow you to invoke \TeX\ by typing a
one-liner like `|tex story|' instead of waiting for the `|**|'; similarly,
`|tex \relax|' works for Experiment~1, and `|tex &plain story|' loads the
plain format before inputting the |story| file.  You might want to try
this, to see if it works on your computer, or you might ask somebody if
there's a similar shortcut.

As \TeX\ begins to read your story file, it types `|(story.tex|', possibly
with a version number for more precise identification, depending on your
local operating system. Then it types `|[1]|', meaning that page~1 is done;
and `|)|', meaning that the file has been entirely input.

\TeX\ will now prompt you with `|*|', because the file did not contain
`^|\end|'. Enter |\end| into the computer now, and you should get a file
|story.dvi| containing a typeset version of Thor's story. As in Experiment~1,
you can proceed to convert |story.dvi| into hardcopy; go ahead and do that now.
The typeset output won't be shown here, but you can see the results by
doing the experiment personally. Please do so before reading on.

\exercise Statistics show that only 7.43 of 10 people who read this manual
actually type the |story.tex| file as recommended, but that those people
learn \TeX\ best. So why don't you join them?
\answer Laziness and/or obstinacy.

\exercise Look closely at the output of Experiment~2, and compare it to
|story.tex|\thinspace: If you followed the instructions carefully, you
will notice a typographical error. What is it, and why did it sneak in?
\answer There's an unwanted space after `called---', because (as the book
says) \TeX\ treats the end of a line as if it were a blank space. That
blank space is usually what you want, except when a line ends with a
hyphen or a dash; so you should {\sc WATCH OUT} for lines that end with
hyphens or dashes.

With Experiment 2 under your belt, you know how to make a document from a
file. The remaining experiments in this chapter are intended to help you
cope with the inevitable anomalies that you will run into later; we will
intentionally do things that will cause \TeX\ to ``squeak.''

But before going on, it's best to fix the error revealed by the previous
output (see exercise 6.2): Line~13 of the |story.tex| file should be changed to
\begintt
he preferred to be called---% error has been fixed!
\endtt
The `|%|' sign here ^^{percent} is a feature of plain \TeX\ that we haven't
discussed before: It effectively terminates a line of your input file,
without introducing the blank space that \TeX\ ordinarily inserts when
moving to the next line of input. Furthermore, \TeX\ ignores everything
that you type following a |%|, up to the end of that line in the file;
you can therefore put ^{comments} into your manuscript, knowing that the
comments are for your eyes only.

Experiment 3 will be to make \TeX\ work harder, by asking it to set the
story in narrower and narrower columns. Here's how: After starting the
program, type
\begintt
\hsize=4in \input story
\endtt
in response to the `|**|'. This means, ``Set the story in a 4-inch column.''
More precisely, ^|\hsize| is a primitive of \TeX\ that specifies the
horizontal size, i.e., the width of each normal line in the output when a
paragraph is being typeset; and ^|\input| is a primitive that causes \TeX\
to read the specified file. Thus, you are instructing the machine to
change the normal setting of\/ |\hsize| that was defined by plain \TeX, and
then to process |story.tex| under this modification.

\TeX\ should respond by typing something like `|(story.tex [1])|' as
before, followed by `|*|'. Now you should type
\begintt
\hsize=3in \input story
\endtt
and, after \TeX\ says `|(story.tex [2])|' asking for more, type three more lines
\begintt
\hsize=2.5in \input story
\hsize=2in \input story
\end
\endtt
to complete this four-page experiment.

Don't be alarmed when \TeX\ screams `|Overfull| |\hbox|' several times as
it works at the 2-inch size; that's what was supposed to go wrong during
Experiment~3. There simply is no good way to break the given
paragraphs into lines that are exactly two inches wide, without making
the spaces between words come out too large or too small. Plain \TeX\
has been set up to ensure rather strict tolerances on all of the lines it
produces:
\begindisplay
\hbox spread-1em{You don't get spaces between words narrower than this,\ and}\cr
\hbox spread+1.679895em{you don't get spaces between words wider than this.}\cr
\enddisplay
If there's no way to meet these restrictions, you get an ^{overfull box}.
And with the overfull box you also get (1)~a warning message, printed
on your terminal, and (2)~a big black bar inserted at the right of the
offending box, in your output. \ (Look at page~4 of the output from
Experiment~3; the overfull boxes should stick out like sore thumbs.
On the other hand, pages 1--3 should be perfect.)

Of course you don't want overfull boxes in your output, so \TeX\ provides
several ways to remove them; that will be the subject of our Experiment~4.
But first let's look more closely at the results of Experiment~3, since
\TeX\ reported some potentially valuable information when it was forced
to make those boxes too full; you should learn how to read this data:
\begintt
Overfull \hbox (0.98807pt too wide) in paragraph at lines 7--11
\tenrm tant galaxy called []O^^?o^^Xc, there lived||
Overfull \hbox (0.4325pt too wide) in paragraph at lines 7--11
\tenrm a com-puter named R. J. Drof-nats. ||
Overfull \hbox (5.32132pt too wide) in paragraph at lines 12--16
\tenrm he pre-ferred to be called---was hap-||
\endtt
Each overfull box is correlated with its location in your input file
(e.g., the first two were generated when processing the paragraph on
lines 7--11 of |story.tex|), and you also learn by how much the copy
sticks out (e.g., 0.98807 points).

Notice that \TeX\ also shows the contents of the overfull boxes in
abbreviated form. For example, the last one has the words `he preferred
to be called---was hap-', set in font |\tenrm| (10-point roman type);
the first one has a somewhat curious rendering of `\"O\"o\c c', because the
accents appear in strange places within that font. In general, when you
see `^|[]|' in one of these messages, it stands either for
the paragraph indentation or for some sort of complex construction;
in this particular case it stands for an umlaut that has been raised
up to cover an `O'.

\dangerexercise Can you explain the `\|' that appears after
`|lived|' in that message?
\answer It represents the heavy bar that shows up in
your output. \ (This bar wouldn't be present if\/ ^|\overfullrule| had been
set to |0pt|, nor is it present in an underfull box.)

\ddangerexercise Why is there a space before the `\|' in `|Drof-nats.
|\|'\thinspace?
\answer This is the ^|\parfillskip| space that ends the paragraph.
In plain \TeX\ the parfillskip is zero when the last line of the paragraph
is full; hence no space actually appears before the rule in the output
of Experiment~3. But all hskips show up as spaces in an overfull box
message, even if they're zero.

You don't have to take out pencil and paper in order to write down the
overfull box messages that you get before they disappear from view, since
\TeX\ always writes a ``^{transcript}'' or ``^{log file}'' that records what
happened during each session. For example, you should now have a file
called |story.log| containing the transcript of Experiment~3, as well
as a file called |texput.log| containing the transcript of Experiment~1. \
(The transcript of Experiment~2 was probably overwritten when you did
number~3.) \ Take a look at |story.log| now; you will see that the overfull
box messages are accompanied not only by the abbreviated box contents,
but also by some strange-looking data about hboxes and glue and kerns and
such things. This data gives a precise description of what's in that
overfull box; \TeX\ wizards will find such
listings important, if they are called upon to diagnose some mysterious
error, and you too may want to understand \TeX's internal code some day.

The abbreviated forms of overfull boxes show the hyphenations that
\TeX\ tried before it resorted to overfilling. The ^{hyphenation} algorithm,
which is described in Appendix~H\null, is excellent but not perfect; for
example, you can see from the messages in |story.log| that \TeX\ finds the
hyphen in `pre-ferred', and it can even hyphenate `Drof-nats'. Yet it
discovers no hyphen in `galaxy', and every once in
a~while an overfull box problem can be cured simply by giving \TeX\ a hint
about how to hyphenate some word more completely. \ (We will see later that
there are two ways to do this, either by inserting ^{discretionary hyphens}
each time as in `\hbox{|gal\-axy|}', or by saying
`\hbox{|\hyphenation{gal-axy}|}' once at the beginning of your manuscript.)

In the present example, hyphenation is not a problem, since \TeX\ found
and tried all the hyphens that could possibly have helped. The only way to
get rid of the overfull boxes is to change the tolerance, i.e., to allow
wider spaces between words. Indeed, the tolerance that plain \TeX\ uses
for wide lines is completely inappropriate for 2-inch columns; such narrow
columns simply can't be achieved without loosening the constraints, unless
you rewrite the copy to fit.

\TeX\ assigns a numerical value called ``^{badness}'' to each line that
it sets, in order to assess the quality of the spacing. The exact rules
for badness are different for different fonts, and they will be discussed
in Chapter~14; but here is the way badness works for the roman font
of plain \TeX:
\begindisplay \hbadness10000
\hbox spread-.666667em{The badness of this line is 100.}&
  \quad(very tight)\cr
\hbox spread-.333333em{The badness of this line is 12.}&
  \quad(somewhat tight)\cr
\hbox{The badness of this line is 0.}&
  \quad(perfect)\cr
\hbox spread.5em{The badness of this line is 12.}&
  \quad(somewhat loose)\cr
%\hbox spread 1em{The badness of this line is 100.}&
% \quad(loose)\cr % then "looser"
\hbox spread 1.259921em{The badness of this line is 200.}&
  \quad(loose)\cr
%\hbox spread 1.713em{The badness of this line is 500.}&
% \quad(bad)\cr % then "worse"
\hbox spread 2.155em{The badness of this line is 1000.}&
  \quad(bad)\cr
\hbox spread 3.684em{The badness of this line is 5000.}& % actually 4995!
  \quad(awful)\cr
\enddisplay
Plain \TeX\ normally stipulates that no line's badness should exceed 200;
but in our case, the task would be impossible since
\begindisplay \hbadness 10000
`\hbox to 2in{tant galaxy called \"O\"o\c c, there}'\hskip 3em
  has badness 1521;\cr
`\hbox to 2in{he preferred to be called---was}'\hskip 3em
  has badness 568.\cr
\enddisplay
So we turn now to Experiment~4, in which spacing variations that are
more appropriate to narrow columns will be used.

Run \TeX\ again, and begin this time by saying
\begintt
\hsize=2in \tolerance=1600 \input story
\endtt
so that lines with badness up to 1600 will be tolerated. Hurray! There are
^^|\tolerance|
no overfull boxes this time. \ (But you do get a message about an {\sl
underfull\/} box, since \TeX\ reports all boxes whose badness exceeds
a certain threshold called ^|\hbadness|; plain \TeX\ sets |\hbadness=1000|.) \
^^{underfull box}
Now make \TeX\ work still harder by trying
\begintt
\hsize=1.5in \input story
\endtt
(thus leaving the tolerance at 1600 but making the ^{column width} still
^^{measure, see hsize}
skimpier). Alas, overfull boxes return; so try typing
\begintt
\tolerance=10000 \input story
\endtt
in order to see what happens. \TeX\ treats 10000 as if it were ``infinite''
tolerance, allowing arbitrarily wide space; thus, a tolerance of 10000 will
{\sl never\/} produce an overfull box, unless something strange occurs like
an unhyphenatable word that is wider than the column itself.

The underfull box that \TeX\ produces in the 1.5-inch case is really bad;
with such narrow limits, an occasional wide space is unavoidable. But try
\begintt
\raggedright \input story
\endtt
for a change. \ ^^|\raggedright|(This tells \TeX\ not to worry about keeping
the right margin straight, and to keep the spacing uniform within each line.) \
Finally, type
\begintt
\hsize=.75in \input story
\endtt
followed by `|\end|', to complete Experiment 4. This makes the columns
almost impossibly narrow.

\danger The output from this experiment will give you some feeling for the
problem of ^{breaking a paragraph} into approximately equal lines. When the
lines are relatively wide, \TeX\ will almost always find a good solution.
But otherwise you will have to figure out some compromise, and several
options are possible. Suppose you want to ensure that no lines have
badness exceeding~500. Then you could set |\tolerance| to some high
number, and |\hbadness=500|; \TeX\ would not produce overfull boxes, but
it would warn you about the underfull ones. Or you could set
|\tolerance=500|; then \TeX\ might produce overfull boxes. If you really
want to take corrective action, the second alternative is better, because
you can look at an overfull box to see how much sticks out; it becomes
graphically clear what remedies are possible.  On the other hand, if you
don't have time to fix bad spacing---if you just want to know how bad it
is---then the first alternative is better, although it may require more
computer time.

\dangerexercise When |\raggedright| has been specified, badness reflects
the amount of space at the right margin, instead of the spacing between
words. Devise an experiment by which you can easily determine what
badness \TeX\ assigns to each line, when the |story| is set ragged-right
in 1.5-inch columns.
\answer Run \TeX\ with \hbox{|\hsize=1.5in|} \hbox{|\tolerance=10000|}
\hbox{|\raggedright|} \hbox{|\hbadness=-1|} and then |\input story|. \TeX\ will
report the badness of all lines (except the final lines of paragraphs, where
fill glue makes the badness zero).

\danger A parameter called ^|\hfuzz| allows you to ignore boxes that are only
slightly overfull. For example, if you say |\hfuzz=1pt|, a box must stick
out more than one point before it is considered erroneous. Plain \TeX\
sets |\hfuzz=0.1pt|.

\ddangerexercise Inspection of the output from Experiment~4, especially
page~3, shows that with narrow columns it would be better to allow white
space to appear before and after a dash, whenever other spaces in the
same line are being stretched. Define a ^|\dash| macro that does this.
\answer |\def\extraspace{\nobreak \hskip 0pt plus .15em\relax}|\parbreak
|\def\dash{\unskip\extraspace---\extraspace}|\par\nobreak\smallskip\noindent
(If you try this with the story at 2-inch and 1.5-inch sizes, you will
notice a substantial improvement. The |\unskip| allows people to leave a
space before typing |\dash|.  \TeX\ will try to hyphenate before |\dash|,
but not before `|---|'; cf.\ Appendix~H\null. The ^|\relax| at the end of
|\extraspace| is a precaution in case the next word is `|minus|'.)

You were warned that this is a long chapter. But take heart: There's only
one more experiment to do, and then you will know enough about \TeX\ to
run it fearlessly by yourself forever after. The only thing you are still
missing is some information about how to cope with ^{error messages}---i.e.,
not just with warnings about things like overfull boxes, but with cases
where \TeX\ actually stops and asks you what to do next.

Error messages can be terrifying when you aren't prepared for them;
but they can be fun when you have the right attitude. Just remember that
you really haven't hurt the computer's feelings, and that nobody will
hold the errors against you. Then you'll find that running \TeX\ might
actually be a creative experience instead of something to dread.

The first step in Experiment 5 is to plant two intentional mistakes in the
|story.tex| file. Change line~3 to
\begintt
\centerline{\bf A SHORT \ERROR STORY}
\endtt
and change `|\vskip|' to `|\vship|' on line~2.

Now run \TeX\ again; but instead of `|story|' type `|sorry|'. The computer
should respond by saying that it can't find file |sorry.tex|, and it will
ask you to try again. Just hit \<return> this time; you'll see
that you had better give the name of a real file. So type `|story|' and
wait for \TeX\ to find one of the {\sl faux pas\/} in that file.

Ah yes, the machine will soon stop,\footnote*{Some installations of \TeX\ do
not allow interaction. In such cases all you can do is look at the error
messages in your log file, where they will appear together with the ``help''
information.} after typing something like this:
\begintt
! Undefined control sequence.
l.2 \vship
           1in
?
\endtt
\write16{\ifnum\pageno=\vshippage
  \else Redefine \string\vshippage to be \the\pageno\fi}%
\TeX\ begins its error messages with `|!|', and it shows what it was
reading at the time of the error by displaying two lines of context. The
top line of the pair (in this case `|\vship|'\thinspace) shows what \TeX\
has looked at so far, and where it came from (`|l.2|', i.e., line number~2);
the bottom line (in this case `|1in|'\thinspace) shows what \TeX\ has yet
to read.

The `^|?|'\ that appears after the context display means that \TeX\ wants
advice about what to do next. If you've never seen an error message before,
or if you've forgotten what sort of response is expected, you can type
`|?|'\ now (go ahead and try it!); \TeX\ will respond as follows:
\begintt
Type <return> to proceed, S to scroll future error messages,
R to run without stopping, Q to run quietly,
I to insert something, E to edit your file,
1 or ... or 9 to ignore the next 1 to 9 tokens of input,
H for help, X to quit.
\endtt
This is your menu of options. You may choose to continue in various ways:

\smallskip\item{1.}
Simply type \<return>. \TeX\ will resume its processing, after
attempting to recover from the error as best it can.

\smallbreak\item{2.} Type `|S|'. \TeX\ will proceed without
pausing for instructions if further errors arise. Subsequent error messages
will flash by on your terminal, possibly faster than you can read them, and
they will appear in your log file where you can scrutinize them at your
leisure. Thus, `|S|' is sort of like typing \<return> to every message.

\smallbreak\item{3.} Type `|R|'. This is like `|S|' but even stronger,
since it tells \TeX\ not to stop for any reason, not even if a file name
can't be found.

\smallbreak\item{4.} Type `|Q|'. This is like `|R|' but even more so,
since it tells \TeX\ not only to proceed without stopping but also to
suppress all further output to your terminal. It is a fast, but somewhat
reckless, way to proceed (intended for running \TeX\ with no operator in
attendance).

\smallbreak\item{5.} Type `|I|', followed by some text that you want to
insert. \TeX\ will read this line of text before encountering what it
would ordinarily see next. Lines inserted in this way are not assumed to
end with a blank space. ^^{inserting text online}
^^{online interaction, see interaction} ^^{interacting with TeX}

\smallbreak\item{6.} Type a small number (less than 100). \TeX\ will
delete this many characters and control sequences from whatever it is
about to read next, and it will pause again to give you another chance to
look things over.  ^^{deleting tokens}

\smallbreak\item{7.} Type `|H|'. This is what you should do now and whenever
you are faced with an error message that you haven't seen for a~while. \TeX\
has two messages built in for each perceived error: a formal one and an
informal one. The formal message is printed first (e.g., `|! Undefined
control sequence.|'\thinspace); the informal one is printed if you request
more help by typing `|H|', and it also appears in your log file if you
are scrolling error messages. The informal message tries to complement the
formal one by explaining what \TeX\ thinks the trouble is, and often
by suggesting a strategy for recouping your losses.^^{help messages}

\smallbreak\item{8.} Type `|X|'. This stands for ``exit.'' It causes \TeX\
to stop working on your job, after putting the finishing touches on your
|log| file and on any pages that have already been output to your |dvi|
file.  The current (incomplete) page will not be output.

\smallbreak\item{9.} Type `|E|'. This is like `|X|', but it also prepares
the computer to edit the file that \TeX\ is currently reading, at the
current position, so that you can conveniently make a change before
trying again.

\smallbreak\noindent
After you type `|H|' (or `|h|', which also works), you'll get a message
that tries to explain that the control sequence just read by \TeX\
(i.e., |\vship|) has never been assigned a meaning, and that you should
either insert the correct control sequence or you should go on as if the
offending one had not appeared.

In this case, therefore, your best bet is to type
\begintt
I\vskip
\endtt
(and \<return>), with no space after the `|I|'; this effectively replaces
|\vship| by |\vskip|. \ (Do it.)

If you had simply typed \<return> instead of
inserting anything, \TeX\ would have gone ahead and read `|1in|', which
it would have regarded as part of a paragraph to be typeset. Alternatively,
you could have typed `|3|'\thinspace; that would have deleted
`|1in|' from \TeX's input. Or you could have typed `|X|' or `|E|' in
order to correct the spelling error in your file. But it's usually
best to try to detect as many errors as you can, each time you run \TeX,
since that increases your productivity while decreasing your computer bills.
Chapter~27 explains more about the art of steering \TeX\ through
troubled text.

\dangerexercise What would have happened if you had typed `|5|' after
the |\vship| error?
\answer \TeX\ would have deleted five tokens: |1|, |i|, |n|, \],
|\centerline|.  (The space was at the end of line~2, the |\centerline| at the
beginning of line~3.)

\danger You can control the level of interaction by giving commands
in your file as well as online: The \TeX\ primitives ^|\scrollmode|,
^|\nonstopmode|, and ^|\batchmode| correspond respectively to typing
`|S|', `|R|', or `|Q|' in response to an error message, and
^|\errorstopmode| puts you back into the normal level of interaction. \
(Such changes are global, whether or not they appear inside a group.) \
Furthermore, many installations have implemented a way to ^{interrupt}
\TeX\ while it is running; such an interruption causes the program to
revert to |\errorstopmode|, after which it pauses and waits for
further instructions.

What happens next in Experiment 5? \TeX\ will hiccup on the other bug that
we planted in the file. This time, however, the error message is more
elaborate, since the context appears on six lines instead of two:
\begintt
! Undefined control sequence.
<argument> \bf A SHORT \ERROR
                              STORY
\centerline #1->\line {\hss #1
                              \hss }
l.3 \centerline{\bf A SHORT \ERROR STORY}
|null
?
\endtt
You get multiline error messages like this when the error is detected
while \TeX\ is processing some higher-level commands---in this case,
while it is trying to carry out |\centerline|, which is not a primitive
operation (it is defined in plain \TeX). At first, such error
messages will appear to be complete nonsense to you, because much of what
you see is low-level \TeX\ code that you never wrote. But you can overcome
this hangup by getting a feeling for the way \TeX\ operates.

First notice that the context information always appears in pairs of lines.
As before, the top line shows what \TeX\ has just read (\thinspace
`|\bf A SHORT \ERROR|'\thinspace), then comes what it is about to read
(\thinspace`|STORY|'\thinspace). The next pair of lines shows the context
of the first two; it indicates what \TeX\ was doing just before it began to
read the others. In this case, we see that \TeX\ has just read `|#1|', which
is a special code that tells the machine to ``read the first ^{argument} that
is governed by the current control sequence''; i.e., ``now read the stuff that
^|\centerline| is supposed to center on a line.'' The definition in Appendix~B
says that |\centerline|, when applied to some text, is supposed to be carried
out by sticking that text in place of the `|#1|' in `|\line{\hss#1\hss}|'.
So \TeX\ is in the midst of this expansion of\/ |\centerline|, as well as being
in the midst of the text that is to be centered.

\looseness-1
The bottom line shows how far \TeX\ has gotten until now in the |story| file.
\ (Actually the bottom line is blank in this example; what appears to be
the bottom line is really the first of two lines of context, and it
indicates that \TeX\ has read everything including the `|}|' in line~3 of
the file.) \ Thus, the context in this error message gives us a glimpse of
how \TeX\ went about its business. First, it saw |\centerline| at the
beginning of line~3. Then it looked at the definition of\/ |\centerline| and
noticed that |\centerline| takes an ``argument,'' i.e., that |\centerline|
applies to the next character or control sequence or group that follows.
So \TeX\ read~on, and filed `|\bf A SHORT \ERROR STORY|' away as the
argument to |\centerline|.  Then it began to read the expansion, as
defined in Appendix~B\null.  When it reached the |#1|, it began to read
the argument it had saved.  And when it reached |\ERROR|, it complained
about an undefined control sequence.

\dangerexercise Why didn't \TeX\ complain about |\ERROR| being undefined
when |\ERROR| was first encountered, i.e., before reading `|STORY}|' on line~3?
\answer A control sequence like |\centerline| might well define a control
sequence like |\ERROR| before telling \TeX\ to look at |#1|. Therefore
\TeX\ doesn't interpret control sequences when it scans an argument.

When you get a multiline error message like this, the best clues about the
source of the trouble are usually on the bottom line (since that is what
you typed) and on the top line (since that is what triggered the error
message). Somewhere in there you can usually spot the problem.

Where should you go from here? If you type `|H|' now, you'll just get
the same help message about undefined control sequences that you saw
before. If you respond by typing \<return>, \TeX\ will go
on and finish the run, producing output virtually identical to that in
Experiment~2. In other words, the conventional responses won't teach you
anything new. So type `|E|' now; this terminates the run and prepares
the way for you to fix the erroneous file. \ (On some systems, \TeX\ will
actually start up the standard text editor, and you'll be positioned at
the right place to delete `|\ERROR|'. On other systems, \TeX\ will simply
tell you to edit line~3 of file |story.tex|.) ^^{editing}

When you edit |story.tex| again, you'll notice that line~2 still contains
|\vship|; the fact that you told \TeX\ to insert |\vskip| doesn't mean
that your file has changed in any way. In general, you should correct all
errors in the input file that were spotted by \TeX\ during a run; the
log file provides a handy way to remember what those errors were.

\smallskip
Well, this has indeed been a long chapter, so let's summarize what has
been accomplished. By doing the five experiments you have learned at first
hand (1)~how to get a job printed via \TeX; (2)~how to make a file that
contains a complete \TeX\ manuscript; (3)~how to change the plain \TeX\
format to achieve columns with different widths; and (4)~how to avoid
panic when \TeX\ issues stern~warnings.

So you could now stop reading this book and go on to print a bunch of
documents. It is better, however, to continue bearing with the author
(after perhaps taking another rest), since you're just at the threshold
of being able to do a lot more. And you ought to read Chapter~7
at least, because it warns you about certain symbols that you must not
type unless you want \TeX\ to do something special. While reading the
remaining chapters it will, of course, be best for you to continue making
trial runs, using experiments of your own design.

\ddanger If you use \TeX\ format packages designed by others, your
error messages may involve many inscrutable two-line levels of macro
context. By setting ^|\errorcontextlines||=0| at the beginning of your file,
you can reduce the amount of information that is reported;
\TeX\ will show only the top and bottom pairs of context lines
together with up to |\errorcontextlines| additional two-line items. \ (If
anything has thereby been omitted, you'll also see `|...|'.) \ Chances
are good that you can spot the source of an error even when most of a
large context has been suppressed; if not, you can say
`|I\errorcontextlines=100| |\oops|' and try again. \ (That will usually
give you an undefined control sequence error and plenty of context.) \
Plain \TeX\ sets |\errorcontextlines=5|.

\endchapter

What we have to learn to do we learn by doing.
\author ^{ARISTOTLE}, {\sl Ethica Nicomachea\/} II (c.~325 B.C.)

\bigskip

He may run who reads.
\author ^{HABAKKUK} 2\thinspace:\thinspace2 (c.~600 B.C.) % RSV
\smallskip
He that runs may read.
\author WILLIAM ^{COWPER}, {\sl Tirocinium\/} (1785)

\eject
\beginchapter Chapter 7. How \TeX\ Reads\\What You Type

We observed in the previous chapter that an input manuscript is expressed
in terms of ``lines,'' but that these lines of input are essentially
independent of the lines of output that will appear on the finished pages.
Thus you can stop typing a line of input at any place that's convenient for
you, as you prepare or edit a file. A few other related rules have also
been mentioned:

\medskip
\item\bull A $\langle\hbox{return}\rangle$ is like a space.

\smallskip
\item\bull Two spaces in a row count as one space.

\smallskip
\item\bull A blank line denotes the end of a paragraph.

\medskip
\noindent Strictly speaking, these rules are contradictory: A blank line
is obtained by typing $\langle\hbox{return}\rangle$ twice in a row,
and this is different from typing two spaces in a row. Some day you might want
to know the {\sl real\/} rules. In this chapter and the next, we shall study
the very first stage in the transition from input to output.

\smallskip
In the first place, it's wise to have a precise idea of what your keyboard
sends to the machine. There are 256 characters that \TeX\ might encounter at
each step, in a file or in a line of text typed directly on your terminal. These
256~characters are classified into 16 categories numbered 0 to 15:
\begindisplay \def\\{\hfill}
\hfil\hidewidth\it Category\hidewidth&\it \qquad Meaning\hidewidth\cr
\noalign{\smallskip}
\\0&Escape character&(|\| in this manual)\cr
\\1&Beginning of group&(|{| in this manual)\cr
\\2&End of group&(|}| in this manual)\cr
\\3&Math shift&(|$| in this manual)\cr
\\4&Alignment tab&(|&| in this manual)\cr
\\5&End of line&(\<return> in this manual)\cr
\\6&Parameter&(|#| in this manual)\cr
\\7&Superscript&(|^| in this manual)\cr
\\8&Subscript&(|_| in this manual)\cr
\\9&Ignored character&(\<null> in this manual)\cr
10&Space&(\] in this manual)\cr
11&Letter&(|A|, \dots, |Z| and |a|, \dots, |z|)\cr
12&Other character&(none of the above or below)\cr
13&Active character&(|~| in this manual)\cr
14&Comment character&(|%| in this manual)\cr
15&Invalid character&(\<delete> in this manual)\cr
\enddisplay
^^{escape character}
^^{begin-group character}
^^{end-group character}
^^{math mode character}
^^{alignment tab}
^^{parameter}
^^{superscript}
^^{subscript}
^^{ignored character}
^^{space}
^^{letter}
^^{other character}
^^{active character}
^^{comment character}
^^{invalid character}
^^{category codes, table}
It's not necessary for you to learn these code numbers; the point is only that
\TeX\ responds to 16~different types of characters. At first this manual led
you to believe that there were just two types---the escape character and the
others---and then you were told about two more types, the grouping
symbols |{| and~|}|. In Chapter~6 you learned two more: |~| and~|%|.
Now you know that there are really~16. This is the whole truth of the
matter; no more types remain to be revealed.  The category code for any
character can be changed at any time, but it is usually wise to stick to a
^^{reserved character} ^^{special character table} ^^\<null> ^^\<delete>
particular scheme.

The main thing to bear in mind is that each \TeX\ format reserves certain
characters for its own special purposes. For example, when you are using plain
\TeX\ format (Appendix~B\null), you need to know that the ten characters
\begintt
\ { } $ & # ^ _ % ~
\endtt
cannot be used in the ordinary way when you are typing;
^^{special characters}
^^{backslash}^^{left brace}^^{right brace}^^{dollar sign}^^{ampersand}
^^{hash mark}^^{hat}^^{underline}^^{percent}^^{tilde}
^^{single-character control sequences}
each of them will cause \TeX\ to do something special, as explained elsewhere
in this book. If you really need these symbols as part of your manuscript,
plain \TeX\ makes it possible for you to type
\begindisplay
|\$| for \$,\qquad |\%| for \%,\qquad |\&| for \&,\qquad
|\#| for \#,\qquad |\_| for \_\thinspace;
\enddisplay
the |\_| symbol is useful for {\it compound\_identifiers\/} in computer
^^{identifiers} ^^{computer programs}
programs. In mathematics formulas you can use |\{| and |\}| for $\{$ and
$\}$, while ^|\backslash| produces a ^{reverse slash}; for example,
\begindisplay
`|$\{a \backslash b\}$|'\quad yields\quad`$\{a\backslash b\}$'.
\enddisplay
Furthermore |\^| produces a circumflex accent (e.g., `|\^e|' yields
`\^e'\thinspace); and |\~| yields a tilde accent (e.g., `|\~n|' yields
`\~n'\thinspace).

\exercise What horrible errors appear in the following sentence?
^^{Procter} ^^{Gamble}
\begintt
Procter & Gamble's stock climbed to $2, a 10% gain.
\endtt
\answer Three forbidden characters were used. One should type
\begintt
Procter \& Gamble's ... \$2, a 10\% gain.
\endtt
(Also the facts are wrong.)

\exercise Can you imagine why the designer of plain \TeX\ decided not
to make `|\\|' the control sequence for reverse slashes?^^{backslash}
\answer Reverse slashes (backslashes) are fairly uncommon in formulas or
text, and |\\| is very easy to type; it was therefore felt best not to
reserve |\\| for such limited use. Typists can define |\\| to be whatever
they want (including |\backslash|).

\danger When \TeX\ reads a line of text from a file, or a line of text that
you entered directly on your terminal, it converts that text into a list of
``^{tokens}.'' A token is either (a)~a single character with an attached
category code, or (b)~a control sequence. For example, if the normal
conventions of plain \TeX\ are in force, the text `|{\hskip 36 pt}|' is
converted into a list of eight tokens:
\begindisplay
|{|$_1$\quad\cstok{hskip}\quad|3|$_{12}$\quad|6|$_{12}$\quad
  \]$_{10}$\quad|p|$_{11}$\quad|t|$_{11}$\quad|}|$_{2}$
\enddisplay
The subscripts here are the category codes, as listed earlier: 1 for
``beginning of group,'' 12 for ``other character,'' and so on. The
\cstok{hskip} doesn't get a subscript, because it represents a control
sequence token instead of a character token. Notice that the space after
|\hskip| does not get into the token list, because it follows a
^{control word}.

\danger It is important to understand the idea of token lists, if you want
to gain a thorough understanding of \TeX, and it is convenient to learn
the concept by thinking of \TeX\ as if it were a living organism. The
individual lines of input in your files are seen only by \TeX's ``eyes''
and ``mouth''; but after that text has been gobbled up, it is sent to
\TeX's ``stomach'' in the form of a token list, and the digestive processes
that do the actual typesetting are based entirely on tokens. As far as the
stomach is concerned, the input flows in as a stream of tokens, somewhat
as if your \TeX\ manuscript had been typed all on one extremely long line.

\danger You should remember two chief things about \TeX's tokens: (1)~A
control sequence is considered to be a single object that is no longer
composed of a sequence of symbols. Therefore long control sequence names
are no harder for \TeX\ to deal with than short ones, after they have been
replaced by tokens. Furthermore, spaces are not ignored after control
sequences inside a token list; the ignore-space rule applies only in an
input file, during the time that strings of characters are being
tokenized.  (2)~Once a category code has been attached to a character
token, the attachment is permanent. For example, if character `|{|' were
suddenly declared to be of category~12 instead of category~1, the
characters `|{|$_1$' already inside token lists of \TeX\ would still
remain of category 1; only newly made lists would contain `|{|$_{12}$'
tokens. In other words, individual characters receive a fixed
interpretation as soon as they have been read from a file, based on the
category they have at the time of reading. Control sequences are
different, since they can change their interpretation at any time.  \TeX's
digestive processes always know exactly what a character token signifies,
because the category code appears in the token itself; but when the
digestive processes encounter a control sequence token, they must look up
the current definition of that control sequence in order to figure out
what it means.

\ddangerexercise Some of the category codes 0 to 15 will never appear as
subscripts in character tokens, because they disappear in \TeX's mouth.
For example, characters of category 0 (escapes) never get to be tokens.
Which categories can actually reach \TeX's stomach?
\answer 1, 2, 3, 4, 6, 7, 8, 10, 11, 12, 13. ^{Active characters} (type 13)
are somewhat special; they behave like control sequences in most cases
(e.g., when you say `^|\let||\x=~|' or `^|\ifx||\x~|'), but they behave like
character tokens when they appear in the token list of\/ ^|\uppercase|
or ^|\lowercase|, and when unexpanded after ^|\if| or ^|\ifcat|.

\ddanger There's a program called ^|INITEX| that is used to install
\TeX, starting from scratch; |INITEX| is like \TeX\ except that it can
do even more things. It can compress ^{hyphenation} patterns into special
tables that facilitate rapid hyphenation, and it can produce ^{format}
files like `|plain.fmt|' from `|plain.tex|'.  But |INITEX| needs extra
space to carry out such tasks, so it generally has less memory available
for typesetting than you would expect to find in a production version of \TeX.

\ddanger When |INITEX| begins, it knows nothing
but \TeX's primitives. All 256~characters are initially of category~12,
except that ^\<return> has category~5,
^\<space> has category~10, ^\<null> has category~9, ^\<delete> has category~15,
the 52 letters |A|$\,\ldots\,$|Z| and |a|$\,\ldots\,$|z| have category~11,
|%| and~|\| have the respective categories 14 and~0. ^^{backslash}^^{percent}
It follows that |INITEX| is initially incapable of carrying out some of
\TeX's primitives that depend on grouping; you can't use |\def| or |\hbox|
until there are characters of categories 1 and~2. The format in
Appendix~B begins with ^|\catcode| commands to provide characters of the
necessary categories; e.g.,
\begintt
\catcode`\{=1
\endtt
assigns category 1 to the |{| symbol. The |\catcode| operation is like
many other primitives of \TeX\ that we shall study later; by modifying
internal quantities like the category codes, you can adapt \TeX\ to a wide
variety of applications.

\ddangerexercise Suppose that the commands
\begintt
\catcode`\<=1 \catcode`\>=2
\endtt
appear near the beginning of a group that begins with `|{|'; these
specifications instruct \TeX\ to treat |<| and |>| as group delimiters.
According to \TeX's rules of locality, the characters |<| and |>| will
revert to their previous categories when the ^{group} ends. But should the
group end with |}| or~with~|>|\thinspace?
\answer It ends with either |>| or |}| or any character of category 2;
then the effects of all |\catcode| definitions within the group are wiped
out, except those that were ^|\global|. \TeX\ doesn't have any built-in
knowledge about how to pair up particular kinds of grouping characters.
New category codes take effect as soon as a |\catcode| assignment has been
digested. For example,
\begintt
{\catcode`\>=2 >
\endtt
is a complete group. But without the space after `|2|' it would not be
complete, since \TeX\ would have read the~`|>|' and converted it to a
token before knowing what category code was being specified; \TeX\ always
reads the token following a constant before evaluating that ^{constant}.

\ddanger Although control sequences are treated as single objects,
\TeX\ does provide a way to break them into lists of character tokens:
If you write ^|\string||\cs|,
where |\cs| is any control sequence, you get the list of characters for that
control sequence's name. For example, |\string\TeX| produces four tokens:
|\|$_{12}$, |T|$_{12}$, |e|$_{12}$, |X|$_{12}$. Each character in this token
list automatically gets category code~12 (``other''),
including the ^{backslash} that |\string| inserts to represent an escape
character.  However, category~10 will be assigned to the character `\]'
(blank ^{space}) if a space character somehow sneaks into the name of a
control sequence.

\ddanger Conversely, you can go from a list of character tokens to a
control sequence by saying `^|\csname|\<tokens>^|\endcsname|'. The tokens
that appear in this construction between |\csname| and |\endcsname| may
include other control sequences, as long as those control sequences
ultimately expand into characters instead of \TeX\ primitives; the final
characters can be of any category, not necessarily letters. For example,
`|\csname TeX\endcsname|' is essentially the same as `|\TeX|'; but
`|\csname\TeX\endcsname|' is illegal, because |\TeX| expands into tokens
containing the ^|\kern| primitive. Furthermore,
`|\csname\string\TeX\endcsname|' will produce the unusual control sequence
`|\\TeX|', i.e., the token \cstok{\char`\\TeX}, which you can't ordinarily
write.

\ddangerexercise Experiment with \TeX\ to see what |\string| does when it
is followed by an ^{active character} like |~|. \ (Active characters behave
like control sequences, but they are not prefixed by an escape.) \ What
is an easy way to conduct such experiments online? What control sequence
could you put after |\string| to~obtain the single character
token~|\|$_{12}$?
\answer If you type `|\message{\string~}|' and `|\message{\string\~}|', \TeX\
responds with `|~|' and `|\~|', respectively. ^^|\message|
To get |\|$_{12}$ from |\string| you therefore need to make backslash an
active character. One way to do this is
\begintt
{\catcode`/=0 \catcode`\\=13 /message{/string\}}
\endtt
(The ``^{null control sequence}'' that you get when there are no
tokens between |\csname| and |\endcsname| is not a solution to this exercise,
because |\string| converts it to `|\csname\endcsname|'. There is, however,
another solution: If \TeX's ^|\escapechar| parameter---which will be
explained in one of the next dangerous bends---is negative or greater
than~255, then `|\string\\|' works.)

\ddangerexercise What tokens does
`|\expandafter\string\csname a\string\   b\endcsname|' produce?
(There are three spaces before the |b|. Chapter~20 explains ^|\expandafter|.)
\answer |\|$_{12}$ |a|$_{12}$ |\|$_{12}$ \]$_{10}$ |b|$_{12}$.

\ddangerexercise When |\csname| is used to define a control sequence for
the first time, that control sequence is made equivalent to |\relax|
until it is redefined. Use this fact to design a macro |\ifundefined#1|
such that, for example,
\begindisplay
|\ifundefined{TeX}|\<true text>|\else|\<false text>|\fi|
\enddisplay
expands to the \<true text> if\/ |\TeX| hasn't previously been defined,
or if\/ |\TeX| has been |\let| equal to |\relax|; it should expand
to the \<false text> otherwise. ^^|\ifundefined|
\answer |\def\ifundefined#1{\expandafter\ifx\csname#1\endcsname\relax}|%
\hfil\break Note that a control sequence like this must be used with care;
it cannot be included in ^{conditional} text, because the |\ifx| will not
be seen when |\ifundefined| isn't expanded.

\ddanger In the examples so far, |\string| has converted control sequences
into lists of tokens that begin with |\|$_{12}$. But this backslash token isn't
really hardwired into \TeX; there's a parameter called ^|\escapechar| that
specifies what character should be used when control sequences are output
as text. The value of\/ |\escapechar| is normally \TeX's internal code for
backslash, but it can be changed if another convention is desired.

\ddanger \TeX\ has two other token-producing operations similar to the
|\string| command. If you write ^|\number|\<number>, you get the decimal
equivalent of the \<number>; and if you write ^|\romannumeral|\<number>,
you get the number expressed in lowercase ^{roman numerals}. For example,
`|\romannumeral24|' produces `|xxiv|', a list of four tokens each having
category~12. The |\number| operation is redundant when it is applied
to an explicit constant (e.g., `|\number24|' produces `|24|'); but it does
suppress leading zeros, and it can also be used with numbers that are in
\TeX's internal registers or parameters. For example, `|\number-0015|'
produces `|-15|'; and if register |\count5| holds the value 316, then
`|\number\count5|' produces `|316|'.

\ddanger The twin operations ^|\uppercase||{|\<token list>|}| and
^|\lowercase||{|\<token list>|}| go through a given token list and convert
all of the character tokens to their ``uppercase'' or ``lowercase''
equivalents. Here's how: Each of the 256 possible characters
has two associated values called the ^|\uccode| and the ^|\lccode|; these values
are changeable just as a |\catcode| is. Conversion to uppercase means
that a character is replaced by its |\uccode| value, unless the |\uccode|
value is zero (when no change is made). Conversion to lowercase is
similar, using the |\lccode|. The category codes aren't changed. When
^|INITEX| begins, all |\uccode| and |\lccode| values are zero except that
the ^{letters} |a| to~|z| and |A| to~|Z| have |\uccode| values |A| to~|Z|
and |\lccode| values |a| to~|z|.

\ddanger \TeX\ performs the  |\uppercase| and |\lowercase| transformations
in its stomach, but the |\string| and |\number| and |\romannumeral|
and |\csname| operations are carried out en route to the stomach (like
macro expansion), as explained in Chapter~20.

\ddangerexercise What token list results from
`|\uppercase{a\lowercase{bC}}|'\thinspace?
\answer First |\uppercase| produces `|A\lowercase{BC}|'; then you get `|Abc|'.

\ddangerexercise \TeX\ has an internal integer parameter called ^|\year| that is
set equal to the current year number at the beginning of every job. Explain how
to use |\year|, together with |\romannumeral| and |\uppercase|, to
print a copyright notice like \year=1986
`\copyright\ \uppercase\expandafter{\romannumeral\year}'
for all jobs run in \number\year.
\answer `\thinspace|\copyright\ \uppercase\expandafter{\romannumeral\year}|%
\thinspace'. \ (This is admittedly tricky; the `^|\expandafter|' expands
the token after the `|{|', not the token after the group.)

\ddangerexercise Define a control sequence |\appendroman| with three parameters
such that |\appendroman#1#2#3| defines control sequence |#1| to
expand to a control sequence whose name is the name of control sequence
|#2| followed by the value of the positive integer |#3| expressed in roman
numerals.  For example, suppose |\count20| equals 30; then
`|\appendroman\a\TeX{\count20}|' should have the same effect as
`|\def\a{\TeXxxx}|'.^^{tricky macros}
\answer (We assume that parameter |#2| is not simply an active character,
and that ^|\escapechar| is between 0 and~255.)
\begintt
\def\gobble#1{} % remove one token
\def\appendroman#1#2#3{\expandafter\def\expandafter#1\expandafter
  {\csname\expandafter\gobble\string#2\romannumeral#3\endcsname}}
\endtt

\endchapter

Some bookes are to bee tasted,
others to bee swallowed,
and some few to bee chewed and disgested.
\author FRANCIS ^{BACON}, {\sl Essayes\/} (1597) % p2 of orig edition

\bigskip

`Tis the good reader that makes the good book.
\author RALPH WALDO ^{EMERSON}, {\sl Society \& Solitude\/} (1870) % Success

\eject
\beginchapter Chapter 8. The Characters\\You Type

A lot of different keyboards are used with \TeX, but few keyboards can
produce 256 different symbols. Furthermore, as we have seen, some of the
characters that you {\sl can\/} type on your ^{keyboard} are reserved for
^^{terminal keyboard}
special purposes like escaping and grouping. Yet when we studied fonts it
was pointed out that there are 256 characters per font. So how can you
refer to the characters that aren't on your keyboard, or that have been
pre\"empted for formatting?

One answer is to use control sequences. For example, the plain format
of Appendix B\null, which defines |%| to be a special kind of symbol so that you
can use it for comments, defines the control sequence |\%| to mean
a ^{percent sign}.

To get access to any character whatsoever, you can type
\begindisplay
|\char|\<number>
\enddisplay
where \<number> is any number from 0 to 255 (optionally followed by a space);
you will get the corresponding character from the current font. That's how
Appendix~B handles |\%|; it defines `|\%|' to be an abbreviation for
`|\char37|', since 37 is the character code for a percent sign.

The codes that \TeX\ uses internally to represent characters are based on
``^{ASCII},'' the American Standard Code for Information Interchange.
^^{internal character codes} ^^{character codes}
Appendix~C gives full details of this code, which assigns numbers to
certain control functions as well as to ordinary letters and punctuation
marks.  For example, ^\<space>${}=32$ and ^\<return>${}=13$.
There are 94~standard visible symbols, and they have been assigned code
numbers from 33 to~126, inclusive.

It turns out that `|b|' is character number 98 in ASCII. So you can
typeset the word |bubble| in a strange way by putting
\begintt
\char98 u\char98\char98 le
\endtt
into your manuscript, if the |b|-key on your keyboard is broken. \
(An optional space is ignored after constants like `|98|'.
Of course you need the |\|, |c|, |h|, |a|, and~|r| keys to type `^|\char|',
so let's hope that they are always working.)

\danger \TeX\ always uses the internal character code of Appendix~C
for the standard ASCII characters,
regardless of what external coding scheme actually appears in the files
being read.  Thus, |b| is 98 inside of \TeX\ even when your computer
normally deals with ^{EBCDIC} or some other non-ASCII scheme; the \TeX\
software has been set up to convert text files to internal code, and to
convert back to the external code when writing text files.
Device-independent (^|dvi|) output files use \TeX's internal code. In
this way, \TeX\ is able to give identical results on all computers.

\danger Character code tables like those in Appendix~C often give the code
numbers in {\sl ^{octal notation}}, i.e., the radix-8 number system, in which
the digits are {\it0},~{\it1}, {\it2}, {\it3}, {\it4}, {\it5}, {\it6},
and~{\it7}.\footnote*{The author of this manual likes to use italic digits
for octal numbers, and typewriter type for hexadecimal numbers, in order
to provide a typographic clue to the underlying radix whenever possible.}
Sometimes {\sl^{hexadecimal notation}\/} is also used, in which case the
digits are |0|,~|1|, |2|, |3|, |4|, |5|, |6|, |7|, |8|, |9|, |A|, |B|, |C|,
|D|, |E|, and~|F|. For example, the octal code for `|b|' is {\it142}, and
its hexadecimal code is |62|. A ^\<number> in \TeX's language can begin
with~a~|'|, in which case it is regarded as octal, or with a |"|, when it is
regarded as hexadecimal. Thus, |\char'142| and |\char"62| are equivalent
to |\char98|. The legitimate character codes in octal notation run from
\oct0 to \oct{377}; in hexadecimal, they run from \hex0 to \hex{FF}.
^^{apostrophe}^^{doublequote}

\danger But \TeX\ actually provides another kind of \<number> that makes it
unnecessary for you to know ASCII at all! The token |`|$_{12}$ (^{left quote}),
when followed by any character token or by any control sequence token
whose name is a single character, stands for \TeX's internal code for the
character in question. For example, |\char`b| and |\char`\b| are also
equivalent to |\char98|. ^^{reverse apostrophe}
If you look in Appendix~B to see how |\%| is defined, you'll notice that
the definition is
\begintt
\def\%{\char`\%}
\endtt
instead of\/ |\char37| as claimed above.

\dangerexercise What would be wrong with |\def\%{\char`%}|?
\answer The |%| would be treated as a comment character, because its
category code is~14; thus, no |%| token or |}| token would get through
to the gullet of \TeX\ where numbers are treated. When a character is
of category 0, 5, 9, 14, or~15, the extra |\| must be used; and the
|\| doesn't hurt, so you can always use it to be safe.

\ddanger The preface to this manual points out that the author
tells little white lies from time to time. Well, if you actually
check Appendix~B you'll find that
\begintt
\chardef\%=`\%
\endtt
is the true definition of\/ |\%|. Since format designers often want to
associate a special character with a special control sequence name, \TeX\
provides the construction `^|\chardef|\<control sequence>|=|\<number>'
for numbers between 0 and 255, as an efficient alternative to
`^|\def|\<control sequence>|{\char|\<number>|}|'.

Although you can use |\char| to access any character in the current
font, you can't use it in the middle of a control sequence. For example,
if you type
\begintt
\\char98
\endtt
\TeX\ reads this as the control sequence |\\| followed by |c|, |h|, |a|,
etc., not as the control sequence |\b|.

You will hardly ever need to use |\char| when typing a manuscript, since
the characters you want will probably be available as predefined control
sequences; |\char| is primarily intended for the designers of book formats
like those in the appendices. But some day you may require a ^{special
symbol}, and you may have to hunt through a font catalog until you find
it. Once you find it, you can use it by simply selecting the appropriate
font and then specifying the character number with |\char|. For example,
the ``^{dangerous bend}'' sign used in this manual appears as character
number~127 of font ^|manfnt|, and that font is selected by the control
sequence ^|\manual|. The macros in Appendix~E therefore display dangerous
bends by saying `|{\manual\char127}|'.

We have observed that the ASCII character set includes only 94 printable
symbols; but \TeX\ works internally with 256 different character codes,
from 0 to 255, each of which is assigned to one of the sixteen categories
described in Chapter~7. If your keyboard has additional symbols, or if it
doesn't have the standard~94, the people who installed your local \TeX\ system
can tell you the correspondence between what you type and the character
number that \TeX\ receives. Some people are fortunate enough to have keys
marked `{\tentex\char'32}' and `{\tentex\char'34}' and `{\tentex\char'35}';
it is possible to install \TeX\ so that it will recognize these handy symbols
and make the typing of mathematics more pleasant. But if you do not have
such keys, you can get by with the control sequences ^|\ne|, ^|\le|,
and ^|\ge|. ^^{not-equal}^^{less-or-equal}^^{greater-or-equal}

\danger \TeX\ has a standard way to refer to the invisible characters of ASCII:
Code~0 can be typed as the sequence of three characters |^^@|, code~1 can
be typed |^^A|, and so on up to code~31, which is |^^_| (see Appendix~C\null).
If the character following |^^| has an internal code between 64 and 127, \TeX\
subtracts 64 from the code; if the code is between 0 and 63, \TeX\
adds~64. Hence code 127 can be typed |^^?|, and
the dangerous bend sign can be obtained by saying
|{\manual^^?}|. However, you must change the category code of character
127 before using it, since this character ordinarily has category~15
(^{invalid}); say, e.g., |\catcode`\^^?=12|.
^^{double hat} ^^{hat hat}
The |^^| notation is different from |\char|, because |^^| combinations are
like single characters; for example, it would not be permissible to say
|\catcode`\char127|, but |^^| symbols can even be used as letters within
control words.

\danger One of the overfull box messages in Chapter 6 illustrates the fact
that \TeX\ sometimes uses the funny |^^| convention in its output:
The umlaut character in that example appears as |^^?|, and the cedilla appears
as~|^^X|, because `\thinspace\"{}\thinspace' and `\char'30' occur in
positions \oct{177} and~\oct{30} of the ^|\tenrm| font.

\danger There's also a special convention in which |^^| is
followed by {\sl two\/} ``lowercase hexadecimal digits,'' |0|--|9| or |a|--|f|.
With this convention, all 256 characters are obtainable in a uniform
way, from |^^00| to |^^ff|. Character 127 is |^^7f|.

\danger Most of the |^^| codes are unimportant except in unusual applications.
But |^^M| is particularly noteworthy because it is code 13, the ASCII
^\<return> that \TeX\ normally assumes to be present
at the right end of every line of ^^{hat hat M}
your input file. By changing the category of~|^^M| you can obtain useful
special effects, as we shall see later.\looseness=-1

\danger The control code |^^I| is also of potential interest, since it's
the ASCII ^\<tab>. Plain \TeX\ makes \<tab> act like a blank space.

\ddanger People who install \TeX\ systems for use with non-American alphabets
can make \TeX\ conform to any desired standard.  For example, suppose
you have a ^{Norwegian keyboard} containing the letter {\tt\ae}, which
^^{Scandinavian letters} ^^{foreign languages}
comes in as code~241 (say). Your local format package should define
|\catcode`|{\tt\ae}|=11|; then you could have control sequences like
|\s|{\tt\ae}|rtrykk|. Your \TeX\ input files could be made readable by
American installations of \TeX\ that don't have your keyboard, by
substituting |^^f1| for character~241. \ (For example, the stated control
sequence would appear as |\s^^f1rtrykk| in the file; your American
friends should also be provided with the format that you used, with its
|\catcode`^^f1=11|.) \ Of course you should also arrange your fonts
so that \TeX's character 241 will print as {\ae}; and you should
change \TeX's hyphenation algorithm so that it will do correct
Norwegian hyphenation. The main point is that such changes are not
extremely difficult; nothing in the design of \TeX\ limits it to the
American alphabet. Fine printing is obtained by fine tuning to the
language or languages being used.
^^{keyboards, non-ASCII}

\ddanger European languages can also be accommodated effectively with
only a limited character set.
For example, let's consider Norwegian again, but suppose that you
want to use a keyboard without an {\tt\ae} character. You can arrange the
^{font metric file} so that \TeX\ will interpret |ae|, |o/|, |aa|, |AE|,
|O/|, and |AA| as ligatures that produce \ae, \o, \aa, \AE, \O, and \AA,
respectively; and you could put the characters \aa\ and \AA\ into positions
128 and~129 of the font.  By setting |\catcode`/=11| you would be able to
use the ligature |o/| in control sequences like `|\ho/yre|'.  \TeX's
hyphenation method is not confused by ligatures; so you could use this
scheme to operate essentially as suggested before, but with two keystrokes
occasionally replacing one. \ (Your typists would have to watch out for
the occasional times when the adjacent characters |aa|, |ae|, and |o/|
should not be treated as ligatures; also, `|\/|' would be a ^{control
word}, not a ^{control symbol}.)

\ddanger The rest of this chapter is devoted to \TeX's reading rules,
which define the conversion from text to tokens. For example, the fact
that \TeX\ ignores spaces after control words is a consequence of
the rules below, which imply among other things that spaces after control
words never become space tokens. The rules are intended to work the
way you would expect them to, so you may not wish to bother reading them;
but when you are communicating with a computer, it is nice to understand
what the machine thinks it is doing, and here's your chance.

\ddanger The input to \TeX\ is a sequence of ``^{lines}.''
Whenever \TeX\ is reading a line of text from a file, or a line of
text that you entered directly on your terminal, the computer's
reading apparatus is in one of three so-called ^{states}:
\begindisplay
\noalign{\vskip1pt}
State $N$&Beginning a new line;\cr
State $M$&Middle of a line;\cr
State $S$&Skipping blanks.\cr
\noalign{\vskip-3pt}
\enddisplay
At the beginning of every line it's in state $N$; but most of the time it's
in state $M$, and after a control word or a space it's in state $S$.
Incidentally, ``states'' are different from the ``^{modes}'' that we will
be studying later; the current {\sl state\/} refers to \TeX's eyes and
mouth as they take in characters of new text, but the current {\sl mode\/}
refers to the condition of \TeX's gastro-intestinal tract. Most of the
things that \TeX\ does when it converts characters to ^{tokens} are independent
of the current state, but there are differences when spaces or end-of-line
characters are detected (categories 10 and 5).

\ddanger \TeX\ deletes any ^\<space> characters (number 32) that occur at the
right end of an input line. Then it inserts a ^\<return> character (number~13)
at the right end of the line, except that it places nothing additional at the
end of a line that you inserted with `|I|'
during ^{error recovery}.  Note that \<return> is considered to be an actual
character that is part of the line; you can obtain special effects by
changing its catcode.

\ddanger If \TeX\ sees an escape character (category 0) in any state, it
scans the entire ^{control sequence} name as follows. (a)~If there are no
more characters in the line, the name is empty (like |\csname\endcsname|).
^^{null control sequence} ^^{csname endcsname}
Otherwise (b)~if the next character is not of category~11 (letter), the
name consists of that single symbol. Otherwise (c)~the name consists of all
letters beginning with the current one and ending just before the first
nonletter, or at the end of the line. This name becomes a control sequence
token. \TeX\ goes into state~$S$ in case~(c), or in case~(b) with respect
to a character of category~10 (space); otherwise \TeX\ goes into state~$M$.

\ddanger If \TeX\ sees a superscript character (category 7) in any state,
and if that character is followed by another identical character, and if
those two equal characters are followed by a character of code
$c<128$, then they
are deleted and 64 is added~to or subtracted from the code~$c$.
\ (Thus, |^^A| is
replaced by a single character whose code is~1, etc., as explained earlier.) \
However, if the two superscript characters are immediately followed by two
of the lowercase hexadecimal digits |0123456789abcdef|, the
four-character sequence is replaced by a single character having the
specified hexadecimal code.
The replacement is carried out also if such a trio or quartet of
characters is encountered during steps (b) or~(c) of the control-sequence-name
scanning procedure described above. After the replacement is made, \TeX\
begins again as if the new character had been present all the time.
If a superscript character is not the first of such a trio or quartet, it is
handled by the following rule.

\ddanger If \TeX\ sees a character of categories 1, 2, 3, 4, 6, 8, 11, 12,
or~13,
or a character of category~7 that is not the first of a special
sequence as just
described, it converts the character to a token by attaching the category
code, and goes into state~$M$. This is the normal case; almost every
nonblank character is handled by this rule.

\ddanger If \TeX\ sees an end-of-line character (category 5), it throws
away any other information that might remain on the current line. Then if
\TeX\ is in state~$N$ (new line), the end-of-line character is converted
to the control sequence token `\cstok{par}' ^^|\par| (end of paragraph); if
\TeX\ is in state~$M$ (mid-line), the end-of-line character is converted
to a token for character~32 (`\]') of category~10 (^{space}); and if \TeX\
is in state~$S$ (skipping blanks), the end-of-line character is simply dropped.

\ddanger If \TeX\ sees a character to be ignored (category~9), it simply
bypasses that character as if it weren't there, and remains in the same state.

\ddanger If \TeX\ sees a character of category~10 (space), the action
depends on the current state. If \TeX\ is in state $N$ or $S$, the
character is simply passed by, and \TeX\ remains in the same state.
Otherwise \TeX\ is in state $M$; the character is converted to a token
of category~10 whose character code is~32, and \TeX\ enters state~$S$.
The character code in a space token is always~32.

\ddanger If \TeX\ sees a comment character (category~14), it throws away that
character and any other information that might remain on the current line.

\ddanger Finally, if \TeX\ sees an invalid character (category~15),
it bypasses that character, prints an error message, and remains in the
same state.

\ddanger If \TeX\ has nothing more to read on the current line, it goes to
the next line and enters state $N$. However, if\/ ^|\endinput| has been
specified for a file being ^|\input|, or if an |\input| file has ended,
\TeX\ returns to whatever it was reading when the |\input| command
was originally given. \ (Further details of\/ |\input| and |\endinput| are
discussed in Chapter~20.)

\ddangerexercise Test your understanding of \TeX's reading rules by answering
the following quickie questions: (a)~What is the difference between
categories 5 and~14? (b)~What is the difference between categories 3
and~4?  (c)~What is the difference between categories 11 and~12?  (d)~Are
spaces ignored after active characters?  (e)~When a line ends with a comment
character like |%|, are spaces ignored at the beginning of the next line?
(f)~Can an ignored character appear in the midst of a control sequence name?
\answer (a)~Both characters terminate the current line; but a character of
category~5 might be converted into \]$_{10}$ or a \cstok{par} token, while
a character of category~14 never produces a token.  (b)~They produce
character tokens stamped with different category numbers.  For example,
|$|$_3$ is not the same token as |$|$_4$, so \TeX's digestive processes
will treat them differently.  (c)~Same as~(b), plus the fact that control
sequence names treat letters differently.  (d)~No. (e)~Yes; characters of
category~10 are ignored at the beginning of every line, since every line
starts in state~$N$. (f)~No.

\ddangerexercise Look again at the error message that appears on page
\vshippage. When \TeX\ reported that |\vship| was an undefined
control sequence, it printed two lines of context, showing that
it was in the midst of reading line~2 of the |story| file. At the
time of that error message, what state was \TeX\ in? What character
was it about to read next?
\answer \TeX\ had just read the control sequence |\vship|, so it
was in state~$S$, and it was just ready to read the space before `|1in|'.
Afterwards it ignored that space, since it was in state~$S$; but if
you had typed |I\obeyspaces| in response to that error message,
you would have seen the space. Incidentally, when \TeX\ prints
the ^{context of an error message}, the bottom pair of lines comes from
a text file, but the other pairs of lines are portions of token lists
that \TeX\ is reading (unless they begin with `|<*>|', when they
represent text entered from the user's terminal, or with `|<insert>|', when they
represent text inserted during ^{error recovery}).

\ddangerexercise Given the category codes of plain \TeX\ format,
what tokens are produced from the input line
`| $x^2$~  \TeX  ^^62^^6|'\thinspace?
\answer |$|$_{3}$ |x|$_{11}$ |^|$_7$ |2|$_{12}$ |$|$_{3}$ |~|$_{13}$ \]$_{10}$
\cstok{TeX} |b|$_{11}$ |v|$_{11}$ \]$_{10}$. The final space comes from the
\<return> placed at the end of the line. Code |^^6| yields |v| only
when not followed by |0|--|9| or |a|--|f|.
The initial space is ignored, because state~$N$
governs the beginning of the line.

\ddangerexercise Consider an input file that contains exactly
three lines; the first line says `|Hi!|', while the other two lines
are completely blank. What tokens are produced when \TeX\ reads
this file, using the category codes of plain \TeX\ format?
\answer |H|$_{11}$ |i|$_{11}$ |!|$_{12}$ \]$_{10}$ \cstok{par}
\cstok{par}. The `\]' comes from the \<return> at the
end of the first line; the second and third lines each contribute
a \cstok{par}.

\ddangerexercise Assume that the category codes of plain \TeX\ are in
force, except that the characters |^^A|, |^^B|, |^^C|, |^^M| belong
respectively to categories 0, 7, 10, and 11. What tokens are produced from
the (rather ridiculous) input line `|^^B^^BM^^A^^B^^C^^M^^@\M|\]'?
(Remember that this line is followed by \<return>, which is
|^^M|; and recall that |^^@| denotes the ^\<null> character, which has
category~9 when |INITEX| begins.)
\answer The two |^^B|'s are not recognized as consecutive superscript
characters, since the first |^^B| is converted to code~2 which doesn't
equal the following character |^|. Hence
the result is seven tokens: |^^B|$_7$ |^^B|$_7$
|M|$_{11}$ \cstok{\^{}\^{}B} \]$_{10}$ |^^M|$_{11}$ \cstok{M\^{}\^{}M}.
The last of these is a control word whose name has two letters.
The \<space> after |\M| is deleted before \TeX\ inserts the \<return> token.

\ddanger The special character inserted at the end of each line needn't
be ^\<return>; \TeX\ actually inserts the current value of an integer
parameter called ^|\endlinechar|, which normally equals~13 but it can
be changed like any other parameter. If the value of\/ |\endlinechar| is
negative or greater than~255, no character is appended, and the effect is
as if every line ends with~|%| (i.e., with a comment character).

\ddanger Since it is possible to change the category codes, \TeX\ might
actually use several different categories for the same character on a single
line. For example, Appendices D and~E contain several ways to coerce \TeX\ to
process text ``^{verbatim},'' so that the author could prepare this manual
without great difficulty. \ (Try to imagine typesetting a \TeX\ manual;
backslashes and other special characters need to switch back and forth
between their normal categories and category~12!) \ Some care is needed to
get the timing right, but you can make \TeX\ behave in a variety of
different ways by judiciously changing the categories. On the other hand,
it is best not to play with the category codes very often, because you must
remember that characters never change their categories once they have become
tokens.  For example, when the arguments to a macro are first scanned,
they are placed into a token list, so their categories are fixed once and
for all at that time.  The author has intentionally kept the category
codes numeric instead of mnemonic, in order to discourage people from
making extensive use of\/ |\catcode| changes except in unusual
circumstances.

\ddangerexercise Appendix B defines ^|\lq| and ^|\rq| to be abbreviations
for |`| and |'| (single left and right quotes, respectively). Explain why
the definitions
\begintt
\chardef\lq=96    \chardef\rq=39
\endtt
would not be as good.
\answer Both alternatives work fine in text; in particular, they combine
as in |\lq\lq| to form ligatures. But the definition in Appendix~B works
also in connection with constants; e.g., |\char\lq\%| and
|\char\rq140| are valid. \ (Incidentally, the construction |\let\lq=`|
would not work with constants, since the quotes in a ^\<number> must
come from character tokens of category~12; after |\let\lq=`| the control
sequence token |\lq| will not expand into a character token, nor {\sl is\/}
it a character token!) ^^|\let| ^^{implicit character}

\endchapter

for life's not a paragraph
\quad
 % he left a blank line here, really
And death i think is no parenthesis
\author e.~e.~^{cummings}, {\sl since feeling is first\/} (1926)

\bigskip

This coded character set is to facilitate
the general interchange of information
among information processing systems,
communication systems, and
associated equipment.
$\ldots$ An 8-bit set was considered
but the need for more than 128 codes
in general applications was not yet evident.
\author ASA SUBCOMMITTEE X3.2, {\sl American Standard\break %
  Code for Information Interchange\/^^{ASCII}} (1963)
% in Communications of the ACM

\eject
\beginchapter Chapter 9. \TeX's\\Roman Fonts

When you're typing a manuscript for \TeX, you need to know what symbols
are available. The plain \TeX\ format of Appendix~B is based on the
Computer Modern fonts, which provide the characters needed to typeset a
wide variety of documents. It's time now to discuss what a person can do
with plain \TeX\ when typing straight text. We've already touched on some of
the slightly subtle things---for example, dashes and quotation marks
were considered in Chapter~2, and certain kinds of accents appeared in the
examples of Chapters 3 and~6. The purpose of this chapter is to give a
more systematic summary of the possibilities, by putting all the facts
together.

Let's begin with the rules for the normal roman font (|\rm| or |\tenrm|);
plain \TeX\ will use this font for everything unless you specify
otherwise. Most of the ordinary symbols that you need are readily available
and you can type them in the ordinary way: There's nothing special about
\begindisplay \openup1pt
the ^{letters} |A| to |Z| and |a| to |z|\cr
the ^{digits} |0| to |9|\cr
common ^{punctuation} marks |: ; ! ? ( ) [ ] ` ' - * / . , @|\cr
\enddisplay
except that \TeX\ recognizes certain combinations as ^{ligatures}:
$$\openup1pt\halign{\indent#\hfil\cr
|ff| yields ff\thinspace;$\!$\quad |fi| yields fi\thinspace;$\!$\quad
|fl| yields fl\thinspace;$\!$\quad
|ffi| yields ffi\thinspace;$\!$\quad |ffl| yields ffl\thinspace;\cr
|``| yields ``\thinspace;\qquad |''| yields ''\thinspace;\qquad
|!||`| yields !`\thinspace;\qquad |?||`| yields ?`\thinspace;\cr
|--| yields --\thinspace;\qquad |---| yields ---\thinspace.\cr}$$
^^{Spanish ligatures}
You can also type |+| and |=|, to get the corresponding
symbols + and~=; but it's much better to use such characters
only in math mode, i.e., enclosed between two |$| signs, since that tells
\TeX\ to insert the proper spacing for mathematics. Math mode is
explained later; for now, it's just a good idea to remember that formulas
and text should be segregated. A non-mathematical hyphen and a non-mathematical
slash should be specified by typing `|-|' and `|/|' outside of mathematics
mode, but subtraction and division should be specified by typing `|-|' and
`|/|' between |$|~signs.
^^{Colon}
^^{Semicolon}
^^{Exclamation point}^^{Shriek, see exclamation point}
^^{Question mark}
^^{Parentheses}
^^{Brackets}
^^{Apostrophe} ^^{Reverse apostrophe}
^^{Hamza, see apostrophe} ^^{Ain, see reverse apostrophe}
^^{Hyphen} ^^{Dash}
^^{Asterisk}
^^{At sign}
^^{Virgule, see slash}
^^{Solidus, see slash}
^^{Shilling sign, see slash}
^^{Slash}
^^{Period} ^^{Full stop, see period}
^^{Comma}
^^{Plus sign}
^^{Equals sign}

The previous paragraph covers 80 of the 94 visible characters of standard
ASCII; so your keyboard probably contains at least 14 more symbols, and
you should learn to watch out for the remaining ones, since they are special.
Four of these are pre\"empted by plain \TeX; if your manuscript requires
the symbols
\begintt
$    #    %    &
\endtt
^^{dollar sign} ^^{sharp sign, see hash mark} ^^{number sign, see hash mark}
^^{hash mark} ^^{percent sign} ^^{ampersand}
you should remember to type them as
\begintt
\$   \#   \%   \&
\endtt
respectively. Plain \TeX\ also reserves the six symbols
\begintt
\    {    }    ^    _    ~
\endtt
^^{backslash} ^^{braces} ^^{curly braces, see braces} ^^{hat, see circumflex}
^^{circumflex} ^^{underline} ^^{tilde}
but you probably don't mind losing these, since they don't appear in
normal copy. Braces and backslashes are available via control sequences
in math mode.

\goodbreak
There are four remaining special characters in the standard ASCII set:
\begintt
"    ||    <    >
\endtt
Again, you don't really want them when you're typesetting text. \ (Double-quote
marks should be replaced either by |``| or by |''|; vertical
lines and relation signs are needed only in math mode.)
^^{double-quote mark} ^^{vertical line, see norm} ^^{norm symbol}
^^{less than sign} ^^{greater than sign}

Scholarly publications in English often refer to other languages, so
plain \TeX\ makes it possible to typeset the most commonly used ^{accents}:
$$\halign{\indent\hbox to 50pt{#\hfil}&\hbox to 35pt{#\hfil}&#\hfil\cr
\it\negthinspace Type&\it to get\cr
\noalign{\smallskip}
|\`o|&\`o&(grave accent)\cr
|\'o|&\'o&(acute accent)\cr
|\^o|&\^o&(circumflex or ``hat'')\cr
|\"o|&\"o&(umlaut or dieresis)\cr
|\~o|&\~o&(tilde or ``squiggle'')\cr
|\=o|&\=o&(macron or ``bar'')\cr
|\.o|&\.o&(dot accent)\cr
|\u o|&\u o&(breve accent)\cr
|\v o|&\v o&(h\'a\v cek or ``check'')\cr
|\H o|&\H o&(long Hungarian umlaut)\cr
|\t oo|&\t oo&(tie-after accent)\cr}$$
^^|\`| ^^{grave accent}
^^|\'| ^^{acute accent}
^^{esc hat} ^^{circumflex accent} ^^{hat accent}
^^|\"| ^^{umlaut accent} ^^{dieresis}
^^{esc tilde} ^^{tilde accent} ^^{squiggle accent}
^^|\=| ^^{macron accent} ^^{bar accent}
^^|\.| ^^{dot accent}
^^|\v| ^^{h\'a\v cek accent} ^^{check accent}
^^|\u| ^^{breve accent}
^^|\H| ^^{Hungarian umlaut}
^^|\t| ^^{tie-after accent}
^^{embellished letters, see accents}
Within the font, such accents are designed to appear at the right height
for the letter `o'; but you can use them over any letter, and \TeX\ will
raise an accent that is supposed to be taller. Notice that spaces are needed
in the last four cases, to separate the control sequences from the letters
that follow. You could, however, type `|\H{o}|' in order to avoid putting a
space in the midst of a word.

\medbreak
Plain \TeX\ also provides three accents that go underneath:
$$\halign{\indent\hbox to 50pt{#\hfil}&\hbox to 35pt{#\hfil}&#\hfil\cr
\it\negthinspace Type&\it to get\cr
\noalign{\smallskip}
|\c o|&\c o&(cedilla accent)\cr
|\d o|&\d o&(dot-under accent)\cr
|\b o|&\b o&(bar-under accent)\cr}$$
^^|\c| ^^{cedilla accent}
^^|\d| ^^{dot-under accent} ^^{emphatics, see dot-under}
^^|\b| ^^{bar-under accent}
And there are a few special letters:
$$\halign{\indent\hbox to 50pt{#\hfil}&\hbox to 35pt{#\hfil}&#\hfil\cr
\it\negthinspace Type&\it to get\cr
\noalign{\smallskip}
|\oe,\OE|&\oe,\thinspace\OE&(French ligature OE)\cr
|\ae,\AE|&\ae,\thinspace\AE&(Latin ligature and Scandinavian letter AE)\cr
|\aa,\AA|&\aa,\thinspace\AA&(Scandinavian A-with-circle)\cr
|\o,\O|&\o,\thinspace\O&(Scandinavian O-with-slash)\cr
|\l,\L|&\l,\thinspace\L&(Polish suppressed-L)\cr
|\ss|&\ss&(German ``es-zet'' or sharp S)\cr}$$
^^{Scandinavian letters} ^^{sharp S} ^^{es-zet} ^^{German} ^^{Polish}
^^{Norwegian} ^^{Danish} ^^{Swedish} ^^{suppressed-L}
^^{diphthongs, see \ae, \oe}
The |\rm| font contains also the ^{dotless letters} `\i' and `\j',
which you can obtain by typing `^|\i|' and `^|\j|'. These are needed because
`i' and `j' should lose their dots when they gain an accent. For example,
the right way to obtain `m\=\i n\u us' is to type \hbox{`|m\=\i n\u us|'}
or `|m\={\i}n\u{u}s|'.

This completes our summary of the |\rm| font. Exactly the same conventions
apply to |\bf|, |\sl|, and |\it|, so you don't have to do things differently
when you're using a different typeface. For example, |\bf\"o| yields
{\bf\"o} and |\it\&| yields {\it\&}.  Isn't that nice?

\danger However, |\tt| is slightly different. You will be glad to know that
|ff|, |fi|, and so on are not treated as ligatures when you're using
^{typewriter type}; nor do you get ligatures from dashes and quote marks.
That's fine, because ordinary dashes and ordinary double-quotes are
appropriate when you're trying to imitate a typewriter. Most of the
accents are available too. But |\H|, |\.|, |\l|, and |\L| cannot be
used---the typewriter font contains other symbols in their place.
Indeed, you are suddenly allowed to type |"|, \|, |<|, and |>|;
^^{doublequote} ^^{vertical line} ^^{less than sign} ^^{greater than sign}
see Appendix~F\null. All of the letters, spaces, and other symbols in
|\tt| have the same width.

\exercise What's the non-naive way to type `na\"\i ve'\thinspace?
\answer |na\"\i ve| or |na{\"\i}ve| or |na\"{\i}ve|.

\exercise List some English words that contain accented letters.
\answer Belov\`ed prot\'eg\'e; r\^ole co\"ordinator; souffl\'es, cr\^epes,
p\^at\'es, etc.

\exercise How would you type `\AE sop's \OE uvres en fran\c cais'\thinspace?
\answer |\AE sop's \OE uvres en fran\c cais|.

\exercise Explain what to type in order to get this sentence:
{\sl Commentarii Academi\ae\ scientiarum imperialis petropolitan\ae\/}
became {\sl Akademi\t\i a Nauk SSSR, Doklady}.
\answer |{\sl Commentarii Academi\ae\ scientiarum imperialis|\hfil\break
|petropolitan\ae\/} became {\sl Akademi\t\i a Nauk SSSR, Doklady}.|

\exercise And how would you specify the names
Ernesto ^{Ces\`aro},
P\'al ^{Erd\H os},
\O ystein ^{Ore},
Stanis\l aw \'Swierczkowski, ^^{Swiercz...}
Serge\u\i\ \t Iur'ev, ^^{Iur'ev}
Mu\d hammad ibn M\^us\^a ^{al-Khw\^arizm\^\i}?
\answer |Ernesto Ces\`aro,
P\'al Erd\H os,
\O ystein Ore,
Stanis\l aw \'Swier%|\break|czkowski,
Serge\u\i\ \t Iur'ev,
Mu\d hammad ibn M\^us\^a al-Khw\^arizm\^\i.|

\dangerexercise Devise a way to typeset {\tt P\'al Erd{\bf\H{\tt o}}s}
in typewriter type.
\answer The proper umlaut is |\H|, which isn't available in |\tt|, so
it's necessary to borrow the accent from another font. For example,
\hbox{|{\tt P\'al Erd{\bf\H{\tt o}}s}|} uses a bold accent, which
is suitably dark.

The following symbols come out looking exactly the same whether you are using
|\rm|, |\sl|, |\bf|, |\it|, or |\tt|:
$$\halign{\indent#\hfil\ &\hfil#\hfil&#\hfil\cr
\it\negthinspace Type&\it to get\cr
\noalign{\smallskip}
|\dag|&\dag&(dagger or obelisk)\cr
|\ddag|&\ddag&(double dagger or diesis)\cr
|\S|&\S&(section number sign)\cr
|\P|&\P&(paragraph sign or pilcrow)\cr}$$
^^{dagger} ^^{double dagger} ^^{obelisk} ^^{obelus, see obelisk} ^^{diesis}
^^{section number sign} ^^{paragraph sign} ^^{pilcrow, see paragraph sign}
(They appear in just one style because plain \TeX\ gets them from the
math symbols font. Lots of other symbols are needed for mathematics;
we shall study them later. See Appendix~B for a few more non-math symbols.)

\exercise In plain \TeX's italic font, the `\$' sign comes out as
`{\it\$}\thinspace'.
^^{dollar sign} ^^{British pound sign} ^^{pound sterling} ^^{sterling}
This gives you a way to refer to pounds sterling, but you might want an
italic dollar sign. Can you think of a way to typeset a reference to
the book {\it Europe on {\sl\$}15.00 a day}\thinspace?
\answer |{\it Europe on {\sl\$}15.00 a day\/}|

\ddanger Appendix B shows that plain \TeX\ handles most of the accents
by using \TeX's ^|\accent| primitive. For example, |\'#1| is equivalent
to |{\accent19 #1}|, where |#1| is the argument being accented.
The general rule is that |\accent|\<number> puts
an accent over the next character; the \<number> tells where that accent
appears in the current font. The accent is assumed to be properly
positioned for a character whose height equals the ^{x-height} of the
current font; taller or shorter characters cause the accent to be raised
or lowered, taking due account of the slantedness of the fonts of accenter
and accentee. The width of the final construction is the width of the
character being accented, regardless of the width of the accent.
Mode-independent commands like font changes may appear between the accent
number and the character to be accented, but grouping operations must not
intervene.  If it turns out that no suitable character is present, the
accent will appear by itself as if you had said |\char|\<number> instead
of\/ |\accent|\<number>.  For example, |\'{}| produces \'{}.

\ddangerexercise Why do you think plain \TeX\ defines |\'#1| to be
`|{\accent19 #1}|' instead of simply letting |\'| be an abbreviation
for `|\accent19 |'\thinspace? \ (Why the extra
braces, and why the argument |#1|?)
\answer The extra braces keep font changes local. An argument makes the
use of\/ |\'| more consistent with the use of other accents like |\d|, which
are manufactured from other characters without using the |\accent|
primitive.

\ddanger It's important to remember that these conventions we have discussed
for accents and special letters are not built into \TeX\ itself; they belong
only to the plain \TeX\ format, which uses the Computer Modern fonts. Quite
different conventions will be appropriate when other fonts are involved;
format designers should provide rules for how to obtain accents and
special characters in their particular systems. Plain \TeX\ works well
enough when accents are infrequent, but the conventions of this chapter
are by no means recommended for large-scale applications of \TeX\ to
other languages. For example, a well-designed \TeX\ font for ^{French}
might well treat accents as ligatures, so that one could |e'crire
de cette manie`re nai"ve en franc/ais| without backslashes. (See the
remarks about Norwegian in Chapter~8.)
^^{foreign languages}

\endchapter

Let's doo't after the high Roman fashion.
\author WILLIAM ^{SHAKESPEARE}, {\sl The Tragedie of %
  Anthony and Cleopatra\/} (1606) % Act IV, Scene 13, line 87

\bigskip

English is a straightforward, frank, honest, open-hearted, no-nonsense language,
which has little truck with such devilish devious devices as accents;
indeed U.S. editors and printers are often thrown into a dither
when a foreign word insinuates itself into the language.
However there is one word on which Americans seem to have closed ranks,
printing it confidently, courageously, and almost invariably
complete with accent---the cheese presented to us as M\"unster.
\smallskip
Unfortunately, ^{Munster} doesn't take an accent.
\author WAVERLEY ^{ROOT}, in the {\sl International Herald Tribune\/} (1982)
  % Tuesday 18 May 82 page 8

\eject
\beginchapter Chapter 10. Dimensions

Sometimes you want to tell \TeX\ how big to make a space, or how wide to
make a line. For example, the short story of Chapter~6 used the instruction
`|\vskip .5cm|' to skip vertically by half a centimeter, and we also
said `|\hsize=4in|' to specify a horizontal size of 4~inches. It's time now
to consider the various ways such ^{dimensions} can be communicated to \TeX.

``^{Points}'' and ``^{picas}'' are the traditional units of measure for
printers and compositors in English-speaking countries, so \TeX\
understands points and picas. \TeX\ also understands inches and metric
units, as well as the continental European versions of points and picas.
Each unit of measure is given a two-letter abbreviation, as follows:
^^{units of measure, table}
$$\halign{\indent\tt#&\quad#\hfil\cr
pt&point (baselines in this manual are $12\pt$ apart)\cr
pc&pica ($\rm1\,pc=12\,pt$)\cr
in&inch ($\rm1\,in=72.27\,pt$)\cr
bp&big point ($\rm72\,bp=1\,in$)\cr
cm&centimeter ($\rm2.54\,cm=1\,in$)\cr
mm&millimeter ($\rm10\,mm=1\,cm$)\cr
dd&didot point ($\rm1157\,dd=1238\,pt$)\cr
cc&cicero ($\rm1\,cc=12\,dd$)\cr
sp&scaled point ($\rm65536\,sp=1\,pt$)\cr}$$
^^|pt|^^{point}
^^|pc|^^{pica}
^^|in|^^{inch}
^^|bp|^^{big point}
^^|cm|^^{centimeter}
^^|mm|^^{millimeter}
^^|dd|^^{didot point}^^{Didot, F. A.}
^^|cc|^^{cicero}
^^|sp|^^{scaled point}
The output of \TeX\ is firmly grounded in the metric system, using the
conversion factors shown here as exact ratios.

\exercise How many points are there in 254 centimeters?
\answer Exactly $7227\pt$.

When you want to express some physical dimension to \TeX, type it as
\begindisplay
\<optional sign>\<number>\<unit of measure>\cr
\noalign{\hbox{or}}
\<optional sign>\<digit string>|.|\<digit string>\<unit of measure>\cr
\enddisplay
where an ^\<optional sign> is either a `|+|' or a `|-|' or nothing at all,
and where a ^\<digit string> consists of zero or more consecutive
decimal digits. The `|.|'\ can also be a `|,|'\null.
For example, here are six typical dimensions:
$$\halign{\indent#\hfil&\hskip 6em#\hfil\cr
|3 in|&|29 pc|\cr
|-.013837in|&|+ 42,1 dd|\cr
|0.mm|&|123456789sp|\cr}$$
A plus sign is redundant, but some people occasionally like extra
redundancy once in a~while. Blank spaces are optional before the signs and the
numbers and the units of measure, and you can also put an optional space
after the dimension; but you should not put spaces within the digits
of a number or between the letters of the unit of measure.

\exercise Arrange those six ``typical dimensions'' into order,
from smallest to largest.
\answer $\rm-.013837\,in$, $\rm0.\,mm$, $\rm+42.1\,dd$, $\rm3\,in$,
$\rm29\,pc$, $\rm123456789\,sp$.
\ (The lines of text in this manual are 29~picas wide.)

\dangerexercise Two of the following three dimensions are legitimate
according to \TeX's rules. Which two are they? What do they mean?
Why is the other one incorrect?
\begintt
'.77pt
"Ccc
-,sp
\endtt
\answer The first is not allowed, since octal notation cannot be used with
a decimal point. The second is, however, legal, since a \<number> can be
hexadecimal according to the rule mentioned in Chapter~8; it means
$\rm12\,cc$, which is $\rm144\,dd\approx154.08124\,pt$. The third is also
accepted, since a \<digit string> can be empty; it is a complicated
way to say $\rm0\,sp$.

\smallskip
The following ``rulers'' have been typeset by \TeX\ so that you can get
some idea of how different units compare to each other. If no distortion
has been introduced during the camera work and printing processes that
have taken place after \TeX\ did its work, these rulers are highly accurate.
$$ \abovedisplayskip 15pt plus 4pt minus 4pt
\belowdisplayskip 15pt plus 4pt minus 4pt
\vbox{
\def\1{\vrule height 0pt depth 2pt}
\def\2{\vrule height 0pt depth 4pt}
\def\3{\vrule height 0pt depth 6pt}
\def\4{\vrule height 0pt depth 8pt}
\def\ruler#1#2#3{\leftline{$\vcenter{\hrule\hbox{\4#1}}\,\,\rm#2\,{#3}$}}
\def\\#1{\hbox to .125in{\hfil#1}}
\def\8{\\\1\\\2\\\1\\\3\\\1\\\2\\\1\\\4}
\ruler{\8\8\8\8}4{in}
\vskip 18pt
\def\\#1{\hbox to 10pt{\hfil#1}}
\def\8{\\\1\\\1\\\1\\\1\\\2\\\1\\\1\\\1\\\1\\\4}
\ruler{\8\8\8}{300}{pt}
\vskip 18pt
\def\\#1{\hbox to 10dd{\hfil#1}}
\def\8{\\\1\\\1\\\1\\\1\\\2\\\1\\\1\\\1\\\1\\\4}
\ruler{\8\8\8}{300}{dd}
\vskip 18pt
\def\\#1{\hbox to 5mm{\hfil#1}}
\def\8{\\\2\\\4}
\ruler{\8\8\8\8\8\8\8\8\8\8}{10}{cm}
\vskip 6pt}$$

\dangerexercise (To be worked after you know about boxes and glue and have
read Chapter~21.) \ Explain how to typeset such a $\rm10\,cm$ ^{ruler},
using \TeX.
\answer {\obeylines|\def\tick#1{\vrule height 0pt depth #1pt}|
|\def\\{\hbox to 1cm{\hfil\tick4\hfil\tick8}}|
|\vbox{\hrule\hbox{\tick8\\\\\\\\\\\\\\\\\\\\}}|
\noindent(You might also try putting ticks at every millimeter, in order %
to see how good your system is; %
some output devices can't handle 101~rules all at once.)}

\danger \TeX\ represents all dimensions internally as an integer multiple
of the tiny units called sp. Since the wavelength of visible light is
approximately $\rm100\,sp$, % in fact: violet=75sp, red=135sp!
rounding errors of a few sp make no difference to the eye.
However, \TeX\ does all of its arithmetic very carefully so that
identical results will be obtained on different computers. Different
implementations of \TeX\ will produce the same line breaks and the same
page breaks when presented with the same document, because the integer
arithmetic will be the same.
^^{machine-independence} ^^{rounding}

\danger The units have been defined here so that precise conversion to~sp
is~efficient on a wide variety of machines. In order to achieve this,
\TeX's ``pt'' has been made slightly larger than the official printer's
point, which was defined to equal exactly $\rm.013837\,in$ by the American
Typefounders Association in~1886 [cf.~National Bureau of Standards
Circular~570 (1956)]. In fact, one classical point is exactly
$.99999999\pt$, so the ``error'' is essentially one part in $10^8$.
This is more than two orders of magnitude less than the amount by which
the inch itself changed during 1959, when it shrank to $\rm2.54\,cm$ from
its former value of $\rm(1/0.3937)\,cm$; so there is no point in worrying
about the difference. The new definition $\rm72.27\,pt=1\,in$ is not only
better for calculation, it~is also easier to remember.

\danger \TeX\ will not deal with dimensions whose absolute value is
$\rm2^{30}\,sp$ or more. In other words, the ^{maximum legal dimension} is
slightly less than $16384\pt$. This is a distance of about 18.892 feet
(5.7583 meters), so it won't cramp your style.

In a language manual like this it is convenient to use ``^{angle brackets}''
in abbreviations for various constructions like \<number> and \<optional
sign> and \<digit string>. Henceforth we shall use the term ^\<dimen> to
stand for a legitimate \TeX\ dimension. For example,
\begindisplay
|\hsize=|\<dimen>
\enddisplay
will be the general way to define the column width that \TeX\ is supposed
to use. The idea is that \<dimen> can be replaced by any quantity like
`|4in|' that satisfies \TeX's grammatical rules for dimensions;
abbreviations in angle brackets make it easy to state such laws of grammar.

When a dimension is zero, you have to specify a unit of measure even
though the unit is irrelevant. Don't just say `|0|'\thinspace; say `|0pt|' or
`|0in|' or something.

\smallbreak
The 10-point size of type that you are now reading is normal in textbooks,
but you probably will often find yourself wanting a larger font. Plain \TeX\
makes it easy to do this by providing {\magnifiedfiverm ^{magnif{}ied
output}.} If you say
\begintt
\magnification=1200
\endtt
at the beginning of your manuscript, everything will be enlarged by 20\%;
i.e., it will come out at 1.2 times the normal size. Similarly,
`|\magnification=2000|' doubles everything; this actually quadruples the area of
each letter, since heights and widths are both doubled. To magnify a
document by the factor $f$, you say ^|\magnification||=|\<number>, where
the \<number> is 1000~times~$f$. This instruction must be given before the
first page of output has been completed. You cannot apply two different
magnifications to the same document.

Magnification has obvious advantages: You'll have less ^{eyestrain} when
you're ^{proofreading}; you can easily make ^{transparencies} ^^{slides}
for lectures; and you can photo-reduce magnified output, in order to minimize
the deficiencies of a ^{low-resolution printer}. Conversely, you might
even want `|\magnification=500|' in order to create a ^{pocket-size}
version of some book. ^^{squint print} But there's a slight catch:
You can't use magnification unless your printing device happens to have the
fonts that you need at the magnification you desire. In other words, you need
to find out what sizes are available before you can magnify. Most
installations of \TeX\ make it possible to print all the fonts of plain
\TeX\ if you magnify by ^|\magstep||0|,~|1|, |2|,~|3|, and perhaps~|4| or
even~|5| (see Chapter~4); but the use of large fonts can be expensive
because a lot of system memory space is often required to store the shapes.

\exercise Try printing the short story of Chapter 6 at 1.2, 1.44, and 1.728
times the normal size. What should you type to get \TeX\ to do this?
\answer For example, say `|\magnification=\magstep1 \input story \end|'
to get magnification 1200; |\magstep2| and |\magstep3| are 1440 and 1728.
Three separate runs are needed, since there can be at most one
magnification per job. The output may look funny if the fonts don't exist
at the stated magnifications.

\danger When you say |\magnification=2000|, an operation like
`|\vskip.5cm|' will actually skip $\rm1.0\,cm$ of space in the final
document. If you want to specify a dimension in terms of the final size,
\TeX\ allows you to say `^|true|' just before |pt|, |pc|, |in|, |bp|,
|cm|, |mm|, |dd|, |cc|, and |sp|.  This unmagnifies the units, so that the
subsequent magnification will cancel out. For example, `|\vskip.5truecm|'
is equivalent to `|\vskip.25cm|' if you have previously said
`|\magnification=2000|'. Plain \TeX\ uses this feature in the
|\magnification| command itself: Appendix~B includes the instruction
\begintt
\hsize = 6.5 true in
\endtt
just after a new magnification has taken effect. This adjusts the line width
so that the material on each page will be $6{1\over2}$ inches wide when it
is finally printed, regardless of the magnification factor.
There will be an inch of margin at both left and right,
assuming that the paper is $8{1\over2}$ inches wide.

\danger If you use no `|true|' dimensions, \TeX's internal computations are not
affected by the presence or absence of magnification; line breaks and page
breaks will be the same, and the ^|dvi| file will change in only two places.
\TeX\ simply tells the printing routine that you want a certain magnification,
and the printing routine will do the actual enlargement when it reads the
|dvi| file.

\dangerexercise Chapter 4 mentions that fonts of different magnifications
can be used in the same job, by loading them `^|at|' different sizes.
Explain what fonts will be used when you give the commands
^^{magnified fonts} ^^|scaled|
\begintt
\magnification=\magstep1
\font\first=cmr10 scaled\magstep1
\font\second=cmr10 at 12truept
\endtt
\answer Magnification is by a factor of 1.2. Since font |\first| is |cmr10|
at $12\pt$, it will be |cmr10| at $14.4\pt$ after magnification;
font |\second| will be |cmr10| at $12\pt$. \ (\TeX\ changes
`|12truept|' into `|10pt|', and the final output magnifies it back to
$12\pt$.)

\ddanger Magnification is actually governed by \TeX's ^|\mag| primitive,
which is an integer parameter that should be positive and at~most~32768.
The value of\/ |\mag| is examined in three cases: (1)~just before the
first page is shipped to the |dvi| file; (2)~when computing a |true|
dimension; (3)~when the |dvi| file is being closed. Alternatively,
some implementations of \TeX\ produce non-|dvi| output; they examine
|\mag| in case~(2) and also when shipping out each page. Since each
document has only one magnification, the value of\/ |\mag| must not change
after it has first been examined.

\danger \TeX\ also recognizes two units of measure that are relative
rather than absolute; i.e., they depend on the current context:
\begindisplay
^|em| is the width of a ``^{quad}'' in the current font;\cr
^|ex| is the ``^{x-height}'' of the current font.\cr
\enddisplay
Each font defines its own em and ex values. In olden days, an ``em'' was
the width of an `M', but this is no longer true; ems are simply arbitrary
units that come with a font, and so are exes.  The Computer Modern fonts
have the property that an em-dash is one em wide, each of the ^{digits} 0
to~9 is half an em wide, and lowercase `x' is one ex high; but these are
not hard-and-fast rules for all fonts.
The |\rm| font (^|cmr10|) of plain \TeX\ has $\rm1\,em=10\,pt$
and $\rm1\,ex\approx4.3\,pt$; the |\bf| font (^|cmbx10|) has
$\rm1\,em=11.5\,pt$ and $\rm1\,ex\approx4.44\,pt$; and the |\tt| font
(^|cmtt10|) has $\rm1\,em=10.5\,pt$ and $\rm1\,ex\approx4.3\,pt$. All of
these are ``10-point'' fonts, yet they have different em and ex values.
It~is generally best to use |em| for horizontal measurements and |ex| for
vertical measurements that depend on the current font.

\danger A \<dimen> can also refer to \TeX's internal registers or parameters.
We shall discuss registers later, and a complete definition of everything that
a ^\<dimen> can be will be given in Chapter~24. For now it will suffice to
give some hints about what is to come:
`|\hsize|' stands for the current horizontal line size,
and `|.5\hsize|' is half that amount;
`|2\wd3|' denotes twice the width of register~|\box3|;
`|-\dimen100|' is the negative of register~|\dimen100|.

\ddanger Notice that the unit names in dimensions
are not preceded by backslashes. The same is true of other so-called
^{keywords} of the \TeX\ language. Keywords can be given in uppercase letters
or in a mixture of upper and lower case; e.g., `|Pt|' is equivalent to `|pt|'.
The category codes of these letters are irrelevant; you may, for example,
be using a |p| of category~12 (other) that was generated by expanding
`|\the\hsize|' as explained in Chapter~20.
\TeX\ gives a special interpretation to keywords only when they
appear in certain very restricted contexts. For example, `|pt|' is a
keyword only when it appears after a number in a \<dimen>;
`|at|' is a keyword only when it appears after the external name of a
font in a |\font| declaration.
Here is a complete list of \TeX's keywords, in case you are wondering about
the full set: |at|, |bp|, |by|, |cc|, |cm|, |dd|, |depth|, |em|, |ex|,
|fil|, |height|, |in|, |l|, |minus|, |mm|, |mu|, |pc|, |plus|,
|pt|, |scaled|, |sp|, |spread|, |to|, |true|, |width|. ^^{reserved words}
\ (See Appendix~I for references to the contexts in which each of these is
recognized as a keyword.)

\endchapter

The methods that have hitherto been taken
to discover the measure of the Roman foot,
will, upon examination, be found so unsatisfactory, that
it is no wonder the learned are not yet agreed on that point.
$\ldots$
9 London inches are equal to 8,447 Paris inches.
\author MATTHEW ^{RAPER}, in {\sl Philosophical Transactions\/} (1760)
% ``An Enquiry into the Measure of the {\sl Roman\/} Foot,''
% {\sl Philos.\ Trans.\ \bf51} (1760), 774--823.

\bigskip

\checkequals\sesame\pageno %
Without the letter U,
units would be nits.
\author ^{SESAME STREET}{^^{Children's Television Workshop}} (1970)

\eject
\beginchapter Chapter 11. Boxes

\TeX\ makes complicated pages by starting with simple individual characters
and putting them together in larger units, and putting these together in still
larger units, and so on. Conceptually, it's a big paste-up job. The \TeX nical
terms used to describe such page construction are {\sl ^{boxes}\/} and
{\sl ^{glue}}.

Boxes in \TeX\ are two-dimensional things with a rectangular shape, having
three associated measurements called {\sl^{height}}, {\sl^{width}}, and
{\sl^{depth}}.  Here is a picture of a typical box, showing its so-called
^{reference point} and ^{baseline}:

{\eightpoint
\setbox0=\hbox{$\uparrow$}
\setbox1=\hbox to \wd0{$\hss\mid\hss$} % with luck, they'll line up
\setbox2=\vbox{\copy0
  \nointerlineskip \kern-.5pt \copy1
  \nointerlineskip \kern-.5pt \copy1
  \moveleft 1em\hbox{height}
  \copy1 \nointerlineskip \kern-.5pt
  \copy1 \nointerlineskip \kern-.5pt
  \hbox{$\downarrow$}
  \kern.2pt}
\setbox3=\vbox{\kern.2pt\copy0
  \moveleft 1em\hbox{depth}
  \hbox{$\downarrow$}
  \kern0pt}
\setbox4=\vtop{\kern-3pt % this cancels the null text above the samplebox
  \hbox{\samplebox{\ht2}{\ht3}{6em}{}%
    \kern-6em
    \raise3pt\hbox to 6em{\hss Baseline\hss}}
  \kern3pt
  \arrows{6em}{width}}
\medskip\indent
\setbox0=\hbox{$\vcenter{}$}% \ht0 is the axis height
\lower\ht0\hbox{Reference point$-$\kern-.2em$\rightarrow$\kern2pt}%
\raise\ht2\box4
\kern1.5em
\raise\ht2\vtop{\kern0pt\box2\nointerlineskip\box3}}

\medskip\noindent
From \TeX's viewpoint, a single character from a font is a box; it's one
of the simplest kinds of boxes. The font designer has decided what the
height, width, and depth of the character are, and what the symbol will
look like when it is in the box; \TeX\ uses these dimensions to paste
boxes together, and ultimately to determine the locations of the reference
points for all characters on a page.  In plain \TeX's |\rm| font (|cmr10|), for
example, the letter `h' has a height of 6.9444 points, a width of 5.5555
points, and a depth of zero; the letter `g' has a height of 4.3055
points, a width of 5 points, and a depth of 1.9444 points. Only certain
special characters like parentheses have height plus depth actually equal
to 10 points, although ^|cmr10| is said to be a ``10-point'' font. You needn't
bother to learn these measurements yourself, but it's good to be aware of
the fact that \TeX\ deals with such information; then you can better
understand what the computer does to your manuscript.

The character shape need not fit inside the boundaries of its box. For example,
some characters that are used to build up larger math symbols like matrix
brackets intentionally protrude a little bit, so that they overlap
properly with the rest of the symbol. Slanted letters frequently extend a
little to the right of the box, as if the box were skewed right at the top
and left at the bottom, keeping its baseline fixed. For example, compare
the letter `g' in the |cmr10| and ^|cmsl10| fonts (|\rm| and |\sl|):
\begindisplay
\vbox to 40pt{\ifproofmode\hrule\vfill
  \hsize=2.5in \baselineskip 6pt \fiverm\noindent
  (A figure will be inserted here; too bad you can't see it now.
  It shows two g's, as claimed.)
  \vfill\hrule\fi}
\enddisplay
In both cases \TeX\ thinks that the box is 5 points wide, so both letters get
exactly the same treatment. \TeX\ doesn't have any idea where the ink will
go---only the output device knows this. But the slanted letters will be
spaced properly in spite of \TeX's lack of knowledge, because the baselines
will match up.

Actually the font designer also tells \TeX\ one other thing, the so-called
{\sl^{italic correction}\/}: A number is specified for each character,
telling roughly how far that character extends to the right of its box
boundary, plus a little to spare. For example,
the italic correction for `g' in |cmr10| is $0.1389\pt$, while in |cmsl10|
it is $0.8565\pt$. Chapter~4 points out that this correction is added to the
normal width if you type `^|\/|' just after the character. You should remember
to use |\/| when shifting from a slanted font to an unslanted one, especially
in cases like
\begintt
the so-called {\sl italic correction\/}:
\endtt
since no space intervenes here to compensate for the loss of slant.

\smallbreak
\TeX\ also deals with another simple kind of box, which might be called
a~``^{black box},'' namely, a rectangle like
`\thinspace \vrule width 4pt height 6pt depth 1.5pt \thinspace'
that is to be entirely filled with ink at printing time. You can specify any
height, width, and depth you like for such boxes---but they had better not have
too much area, or the printer might get upset. \ (Printers generally
prefer white space to black space.)

Usually these black boxes are made very skinny, so that they appear as
horizontal lines or vertical lines. Printers traditionally call such lines
``^{horizontal rules}'' and ``^{vertical rules},'' so the terms \TeX\ uses
to stand for black boxes are ^|\hrule| and ^|\vrule|. Even when the box is
square, as in `\thinspace\bull\thinspace', you must call it either an~|\hrule|
or a~|\vrule|.  We shall discuss the use of ^{rule boxes} in greater
detail later.  \ (See Chapter~21.)

\smallbreak
Everything on a page that has been typeset by \TeX\ is made up of simple
character boxes or rule boxes, pasted together in combination. \TeX\
pastes boxes together in two ways, either {\sl horizontally\/} or {\sl
vertically}.  When \TeX\ builds a ^{horizontal list} of boxes, it lines
them up so that their reference points appear in the same horizontal row;
therefore the baselines of adjacent characters will match up as they
should. Similarly, when \TeX\ builds a ^{vertical list} of boxes, it lines
them up so that their reference points appear in the same vertical column.

% Here are some macros for making blank boxes
\def\dolist{\afterassignment\dodolist\let\next= }
\def\dodolist{\ifx\next\endlist \let\next\relax
  \else \\\let\next\dolist \fi
  \next}
\def\endlist{\endlist}
\def\\{\expandafter\if\space\next\ \else \setbox0=\hbox{\next}\maketypebox\fi}
\def\demobox#1{\setbox0=\hbox{\dolist#1\endlist}%
  \copy0\kern-\wd0\makelightbox}

Let's take a look at what \TeX\ does behind the scenes, by comparing the
computer's methods with what you would do if you were setting metal type
by hand. In the time-tested traditional method, you choose the letters that
you~need out of a type case---the uppercase letters are in the ^{upper
case}---and you put them into a ``^{composing stick}.'' When a line is
complete, you adjust the spacing and transfer the result to the ``chase,''
where it joins the other rows of type. Eventually you lock the type up
tightly by adjusting external wedges called ``quoins.'' This isn't much
different from what \TeX\ does, except that different words are used; when
\TeX\ locks up a line, it creates what is called an ``^{hbox}''
(^{horizontal box}), because the components of the line are pieced
together horizontally. You can give an instruction like
\begintt
\hbox{A line of type.}
\endtt
in a \TeX\ manuscript; this tells the computer to take boxes for the appropriate
letters in the current font and to lock them up in an hbox. As far as \TeX\ is
concerned, the letter `A' is a box
`\thinspace\setbox0\hbox{A}\maketypebox\thinspace'
and the letter `p' is a box
`\thinspace\setbox0\hbox{p}\maketypebox\thinspace'.
So the given instruction causes \TeX\ to form the hbox
\begindisplay
\demobox{A line of type.}
\enddisplay
representing `A line of type.' The hboxes for individual lines of type are
eventually joined together by putting them into a ``^{vbox}'' (^{vertical
box}). For example, you can say
\begintt
\vbox{\hbox{Two lines}\hbox{of type.}}
\endtt
and \TeX\ will convert this into
\begindisplay%
  \setbox0=\vbox{\hbox{\demobox{Two lines}}\hbox{\demobox{of type.}}}
$\vcenter{\hbox{\makelightbox\kern-\wd0\box0}}$\qquad
  i.e.,\qquad$\vcenter{\vbox{\hbox{Two lines}\hbox{of type.}}}$
\enddisplay
The principal difference between \TeX's method and the old way is that metal
types are generally cast so that each character has the same height and
depth; this makes it easy to line them up by hand. \TeX's types have
variable height and depth, because the computer has no trouble lining
characters up by their baselines, and because the extra information about
height and depth helps in the positioning of accents and mathematical
symbols.

Another important difference between \TeX\ setting and hand setting is, of
course, that \TeX\ will choose line divisions automatically; you don't
have to insert ^|\hbox| and ^|\vbox| instructions unless you want to
retain complete control over where each letter goes. On the other hand,
if you do use |\hbox| and |\vbox|, you can make \TeX\ do almost everything
that Ben ^{Franklin} could do in his printer's shop. You're only giving
up the ability to make the letters come out charmingly crooked or badly
inked; for such effects you need to make a new font. \ (And of course you
lose the tactile and olfactory sensations, and the thrill of
doing everything by yourself. \TeX\ will never completely replace the
good~old~ways.)

A page of text like the one you're reading is itself a box, in \TeX's view:
It is a largish box made from a vertical list of smaller boxes representing
the lines of text. Each line of text, in turn, is a box made from a
horizontal list of boxes representing the individual characters. In more
complicated situations, involving mathematical formulas and/or complex
tables, you can have boxes within boxes within boxes $\ldots$ to any level.
But even these complicated situations arise from horizontal or vertical lists
of boxes pasted together in a simple way; all that you and \TeX\ have to
worry about is one list of boxes at a time. In fact, when you're typing
straight text, you don't have to think about boxes at all, since \TeX\ will
automatically take responsibility for assembling the character boxes into
words and the words into lines and the lines into pages. You need to be
aware of the box concept only when you want to do something out of the
ordinary, e.g., when you want to center a heading.

\danger From the standpoint of \TeX's digestive processes, a manuscript
comes in as a sequence of tokens, and the tokens are to be transformed into
a sequence of boxes. Each token of input is essentially an instruction or
a piece of an instruction; for example, the token `|A|$_{11}$' normally means,
``put a character box for the letter |A| at the end of the current hbox,
using the current font''; the token `\cstok{vskip}' normally means, ``skip
vertically in the current vbox by the \<dimen> specified in the
following tokens.''

\danger The height, width, or depth of a box might be negative, in which
case it is a ``^{shadow box}'' that is somewhat hard to draw. \TeX\ doesn't
balk at ^{negative dimensions}; it just does arithmetic as usual. For example,
the combined width of two adjacent boxes is the sum of their widths, whether
or not the widths are positive.  A font designer can declare a character's
width to be negative, in which case the character acts like a ^{backspace}. \
(Languages that read from right to left could be
handled in this way, but only to a limited extent, since \TeX's line-breaking
^^{Hebrew} ^^{Arabic}
algorithm is based on the assumption that words don't have negative widths.)

\danger \TeX\ can raise or lower the individual boxes
in a horizontal list; such adjustments take care of mathematical
subscripts and superscripts, as well as the heights of accents and a few
other things. For example, here is a way to make a box that contains
the \TeX\ logo, putting it into \TeX's internal register |\box0|:
\begintt
\setbox0=\hbox{T\kern-.1667em\lower.5ex\hbox{E}\kern-.125em X}
\endtt
^^|\setbox|
Here `^|\kern||-.1667em|' means to insert blank space of $-.1667$ ems in the
current font, i.e., to back up a bit; and `^|\lower||.5ex|' means that
the box |\hbox{E}| is to be lowered by half of the current x-height, thus
offsetting that box with respect to the others. Instead of
`|\lower.5ex|' one could also say `^|\raise||-.5ex|'. Chapters 12 and~21
discuss the details of how to construct boxes for special effects;
our goal in the present chapter is merely to get a taste of the
possibilities.

\danger \TeX\ will exhibit the contents of any ^{box register}, if you
ask it to. For example, if you type `^|\showbox||0|' after setting
|\box0| to the \TeX\ logo as above, your ^{log file} will contain
the following mumbo jumbo: ^^{TeX logo}
\begintt
\hbox(6.83331+2.15277)x18.6108
.\tenrm T
.\kern -1.66702
.\hbox(6.83331+0.0)x6.80557, shifted 2.15277
..\tenrm E
.\kern -1.25
.\tenrm X
\endtt
^^{diagnostic format} ^^{internal box-and-glue representation} ^^{box displays}
The first line means that |\box0| is an hbox whose height, depth, and width
are respectively $6.83331\pt$, $2.15277\pt$, and $18.6108\pt$.
Subsequent lines beginning with `|.|'\ indicate that they are {\sl inside\/}
of a box. The first thing in this particular box is the letter~|T| in
font |\tenrm|; then comes a kern. The next item is an hbox that contains
only the letter~|E|; this box has the height, depth, and width of an |E|, and
it has been shifted downward by $2.15277\pt$ (thereby accounting for
the depth of the larger box).

\dangerexercise Why are there two dots in the `|..\tenrm E|' line here?
\answer This |E| is inside a box that's inside a box.

\danger Such displays of box contents will be discussed further in
Chapters 12 and~27.
They are used primarily for diagnostic purposes, when you are trying to figure
out exactly what \TeX\ thinks it's doing. The main reason for bringing them
up in the present chapter is simply to provide a glimpse of how \TeX\ represents
boxes in its guts. A computer program doesn't really move boxes around; it
fiddles with lists of representations of boxes.

\dangerexercise By running \TeX, figure out how it actually handles italic
corrections to characters: How are the corrections represented inside a box?
\answer The idea is to construct a box and to look inside. For example,
\begintt
\setbox0=\hbox{\sl g\/} \showbox0
\endtt
reveals that |\/| is implemented by placing a kern after the character.
Further experiment shows that this kern is inserted even when the italic
correction is zero.

\dangerexercise The ``opposite'' of \TeX's logo---namely,
T\kern+.1667em\raise.5ex\hbox{E}\kern+.125em X---is produced by
\begintt
\setbox1=\hbox{T\kern+.1667em\raise.5ex\hbox{E}\kern+.125em X}
\endtt
What would |\showbox1| show now? \ (Try to guess, without running the machine.)
\answer The height, depth, and width of the enclosing box should be just large
enough to enclose all of the contents, so the result is:
\begintt
\hbox(8.98608+0.0)x24.44484
.\tenrm T
.\kern 1.66702
.\hbox(6.83331+0.0)x6.80557, shifted -2.15277
..\tenrm E
.\kern 1.25
.\tenrm X
\endtt
(You probably predicted a height of |8.9861|; \TeX's internal calculations are
in |sp|, not |pt|/100000, so the rounding in the fifth decimal place is not
readily predictable.)

\dangerexercise Why do you think the author of \TeX\ didn't make boxes more
symmetrical between horizontal and vertical, by allowing reference points
to be inside the boundary instead of insisting that the reference point
must appear at the left edge of each box?
\answer No applications of such symmetrical boxes to English-language
printing were apparent; it seemed pointless to carry extra generality
as useless baggage that would rarely if ever be used, merely for the sake of
symmetry. In other words, the author wore a computer science cap instead
of a mathematician's mantle on the day that \TeX's boxes were born.
Time will tell whether or not this was a fundamental error!

\ddangerexercise Construct a |\demobox| macro for use in writing manuals
like this, so that an author can write `|\demobox{Tough exercise.}|'
in order to typeset `\thinspace\demobox{Tough exercise.}\thinspace'.
\answer The following solution is based on a general |\makeblankbox|
macro that prints the edges of a box using rules of given thickness
outside and inside that box; the box dimensions are those of\/ |\box0|.\par
|\def\dolist{\afterassignment\dodolist\let\next= }|\parbreak
|\def\dodolist{\ifx\next\endlist \let\next\relax|\parbreak
|  \else \\\let\next\dolist \fi|\parbreak
|  \next}|\par
|\def\endlist{\endlist}|\par
|\def\hidehrule#1#2{\kern-#1%|\parbreak
|  \hrule height#1 depth#2 \kern-#2 }|\par
|\def\hidevrule#1#2{\kern-#1{\dimen0=#1|\parbreak
|    \advance\dimen0 by#2\vrule width\dimen0}\kern-#2 }|\par
|\def\makeblankbox#1#2{\hbox{\lower\dp0\vbox{\hidehrule{#1}{#2}%|\parbreak
|    \kern-#1 % overlap the rules at the corners|\parbreak
|    \hbox to \wd0{\hidevrule{#1}{#2}%|\parbreak
|      \raise\ht0\vbox to #1{}% set the vrule height|\parbreak
|      \lower\dp0\vtop to #1{}% set the vrule depth|\parbreak
|      \hfil\hidevrule{#2}{#1}}%|\parbreak
|    \kern-#1\hidehrule{#2}{#1}}}}|\par
|\def\maketypebox{\makeblankbox{0pt}{1pt}}|\par
|\def\makelightbox{\makeblankbox{.2pt}{.2pt}}|\par
|\def\\{\if\space\next\ % assume that \next is unexpandable|\parbreak
| \else \setbox0=\hbox{\next}\maketypebox\fi}|\par
|\def\demobox#1{\setbox0=\hbox{\dolist#1\endlist}%|\parbreak
|  \leavevmode\copy0\kern-\wd0\makelightbox}|\par

\def\frac#1/#2{\leavevmode\kern.1em
  \raise.5ex\hbox{\the\scriptfont0 #1}\kern-.1em
  /\kern-.15em\lower.25ex\hbox{\the\scriptfont0 #2}}

\ddangerexercise Construct a |\frac| macro such that `|\frac1/2|' yields
`\frac1/2'. \checkequals\fracexno\exno
\answer |\def\frac#1/#2{\leavevmode\kern.1em|\parbreak
|\raise.5ex\hbox{\the\scriptfont0 #1}\kern-.1em|\parbreak
|/\kern-.15em\lower.25ex\hbox{\the\scriptfont0 #2}}|

\endchapter

I have several boxes in my memory
in which I will keep them all very safe,
% he's talking about "instructions"
there shall not a one of them be lost.
\author IZAAK ^{WALTON}, {\sl The Compleat Angler\/} (1653) % beginning Chap12
% in 1654 and subsequent editions, this quote comes in Chap17
% the 1653 spelling agrees with 20th century conventions in this passage!

\bigskip

How very little does the amateur, dwelling at home at ease,
comprehend the labours and perils of the author.
\author R. L. ^{STEVENSON} and L. ^{OSBOURNE}, {\sl The Wrong Box\/} (1889)

\eject
\beginchapter Chapter 12. Glue

But there's more to the story than just boxes: There's also some magic mortar
called {\sl ^{glue}\/} that
\TeX\ uses to paste boxes together. For example, there is a little space between
the lines of text in this manual; it has been calculated so that the baselines
of consecutive lines within a paragraph are exactly 12~points apart. And
there is space between words too; such space is not an ``empty'' box, it is
part of the glue between boxes. This glue can stretch or shrink so that the
right-hand margin of each page comes out looking straight.
^^{leading, see baselineskip} ^^{skipping space, see glue}

When \TeX\ makes a large box from a horizontal or vertical list of smaller
boxes, there often is glue between the smaller boxes. Glue has three
attributes, namely its natural {\sl space}, its ability to {\sl ^{stretch}}, and
its ability to {\sl ^{shrink}}.

In order to understand how this works, consider the following example of
four boxes in a horizontal list separated by three globs of glue:
\begindisplay\eightpoint
\vbox{
    \hbox{\samplebox{7mm}{8mm}{5\varunit}{width 5}%
      \sampleglue{9\varunit}{space 9\cr stretch 3\cr shrink 1}%
      \samplebox{3mm}{2mm}{6\varunit}{width 6}%
      \sampleglue{9\varunit}{space 9\cr stretch 6\cr shrink 2}%
      \samplebox{8mm}{3mm}{3\varunit}{width 3}%
      \sampleglue{12\varunit}{space 12\cr stretch 0\cr shrink 0}%
      \samplebox{4mm}{7mm}{8\varunit}{width 8}}
    \kern6pt
    \arrows{52\varunit}{width 52}}
\enddisplay
The first glue element has 9 units of space, 3 of stretch, and 1 of shrink;
the next one also has 9 units of space, but 6 units of stretch and 2 of
shrink; the last one has 12 units of space, but it is unable to stretch
or to shrink, so it will remain 12 units of space no matter what.

The total width of boxes and glue in this example, considering only the
space components of the glue, is $5+9+6+9+3+12+8=52$ units. This is called
the {\sl ^{natural width}\/} of the horizontal list; it's the preferred way to
paste the boxes together. Suppose, however, that \TeX\ is told to make the
horizontal list into a box that is 58~units wide; then the glue has to
stretch by 6~units. Well, there are $3+6+0=9$ units of stretchability present,
so \TeX\ multiplies each unit of stretchability by 6/9 in order to obtain the
extra 6~units needed. The first glob of glue becomes $9+(6/9)\times3=11$
units wide, the next becomes $9+(6/9)\times6=13$ units wide, the last remains
12 units wide, and we obtain the desired box looking like this:
\begindisplay\eightpoint
\vbox{\kern-3pt
    \hbox{\samplebox{7mm}{8mm}{5\varunit}{}%
      \sampleglue{11\varunit}{$9+2$}%
      \samplebox{3mm}{2mm}{6\varunit}{}%
      \sampleglue{13\varunit}{$9+4$}%
      \samplebox{8mm}{3mm}{3\varunit}{}%
      \sampleglue{12\varunit}{$12+0$}%
      \samplebox{4mm}{7mm}{8\varunit}{}}
    \kern6pt
    \arrows{58\varunit}{width 58}}
\enddisplay

On the other hand, if \TeX\ is supposed to make a box 51 units wide from the
given list, it is necessary for the glue to shrink by a total of one unit. There
are three units of shrinkability present, so the first glob of glue would
shrink by 1/3 and the second by 2/3.

\smallbreak
The process of determining glue thickness when a box is being made from a
horizontal or vertical list is called {\sl ^{setting the glue}}.  Once glue has
been set, it becomes rigid; it won't stretch or shrink any more, and the
resulting box is essentially indecomposable.

Glue will never shrink more than its stated shrinkability. For example,
the first glob of glue in our illustration will never be allowed to become
narrower than 8 units wide, and \TeX\ will never shrink the given
horizontal list to make its total width less than 49 units.  But glue is
allowed to stretch arbitrarily far, whenever it has a positive stretch
component.

\exercise How wide would the glue globs be if the horizontal list in the
illustration were to be made 100 units wide?
\answer $9+16$ units, $9+32$ units, $12+0$ units. \ (But \TeX\ would
consider so much stretching to be ``infinitely bad.'')

Once you understand \TeX's concept of glue, you may well decide that it
was misnamed; real glue doesn't stretch or shrink in such ways, nor does it
contribute much space between boxes that it welds together. Another word
like ``spring'' would be much closer to the essential idea, since ^{springs}
have a natural width, and since different springs compress and expand at
different rates under tension. But whenever the author has suggested
changing \TeX's terminology, numerous people have said that they like the
word ``glue'' in spite of its inappropriateness; so the original name has
stuck.

\danger \TeX\ is somewhat reluctant to stretch glue more than the stated
stretchability; therefore you can decide how big to make each aspect of the
glue by using the following rules: \ (a)~The natural
glue space should be the amount of space that looks best. \ (b)~The glue
stretch should be the maximum amount of space that can be added to the natural
spacing before the layout begins to look bad. \ (c)~The glue shrink should
be the maximum amount of space that can be subtracted from the natural spacing
before the layout begins to look bad.

In most cases the designer of a book layout will have specified all the kinds
of glue that are to be used, so a typist will not need to decide how big any
glue attributes should be. For example, users of the plain \TeX\ format of
Appendix~B can type `|\smallskip|' when they want a little extra ^{space
between paragraphs}; a ^|\smallskip| turns out to be $3\pt$ worth of
vertical glue that can stretch or shrink by an additional~$1\pt$. Here is
a |\smallskip|: \smallskip

\noindent
Instead of sprinkling various amounts of glue throughout a manuscript,
expressing each of them explicitly in terms of points, you will find it
much better to explain your intentions more clearly by typing something
like `|\smallskip|' when you want abnormal spacing. The definition of\/
|\smallskip| can readily be changed later, in case you want such spaces to
be smaller or larger.  Plain \TeX\ also provides you with `^|\medskip|',
which is worth two smallskips, and `^|\bigskip|', which is worth two medskips.

\danger A plain \TeX\ |\medskip| appears before and after each
``^{dangerous bend}'' section of this manual, so you have already seen
numerous examples of such spacing before you knew what it was called.
Vertical glue is created by writing `|\vskip|\<glue>', where ^\<glue> is
any glue specification. The usual way to specify \<glue> to \TeX\ is
\begindisplay
\<dimen> |plus|\<dimen> |minus|\<dimen>
\enddisplay
where the `|plus|\<dimen>' and `|minus|^\<dimen>' are optional and assumed
to be zero if not present; `^|plus|' introduces the amount of
stretchability, `^|minus|' introduces the amount of shrinkability.
For example, Appendix~B defines |\medskip| to be an abbreviation for
`|\vskip6pt plus2pt minus2pt|'. The normal-space component of glue must
always be given as an explicit \<dimen>, even when it is zero.

\danger Horizontal glue is created in the same way, but with ^|\hskip|
instead of\/ ^|\vskip|. For example, plain \TeX\ defines ^|\enskip| as an
abbreviation for the command `|\hskip.5em\relax|'; this skips horizontally by
one ``^{en},'' i.e., by exactly half of an em in the current font. There is no
stretching or shrinking in an |\enskip|. The control sequence ^|\relax|
after `|.5em|' prevents \TeX\ from thinking that a ^{keyword}
is present, in case the text following |\enskip| just happens to begin
with `|plus|' or `|minus|'.

One of the interesting things that can happen when glue stretches and shrinks
at different rates is that there might be glue with {\sl ^{infinite}\/}
stretchability. For example, consider again the four boxes we had at the
beginning of this chapter, with the same glue as before except that the
glue in the middle can stretch infinitely far. Now the total
stretchability is infinite; and when the line has to grow, all of the
additional space is put into the middle glue. If, for example, a box of
width 58 is desired, the middle glue expands from 9 to~15 units, and the
other spacing remains unchanged.

If such infinitely stretchable glue is placed at the left of a row of
boxes, the effect is to place them ``flush right,'' i.e., to move them
over to the ^^{right justification} ^^{centering} ^^{flush right}
rightmost boundary of the constructed box. And if you take {\sl two\/}
globs of infinitely stretchable glue, putting one at the left and one
at the right, the effect is to {\sl center\/} the list of boxes within
a larger box. This in fact is how the ^|\centerline| instruction works
in plain \TeX\null: It places infinite glue at both ends, then makes a
box whose width is the current value of\/ |\hsize|.

The short story example of Chapter 6 used infinite glue not only for
centering, but also in the ^|\vfill| instruction at the end; `|\vfill|'
essentially means ``skip vertically by zero, but with infinite stretchability.''
In other words, |\vfill| fills up the rest of the current page with
blank space.

\danger \TeX\ actually recognizes several kinds of infinity, some of which
are ``more infinite'' than others. You can say both ^|\vfil| and |\vfill|;
the second is stronger than the first. In other words, if no other
infinite stretchability is present, |\vfil| will expand to fill the remaining
space; but if both |\vfil| and |\vfill| are present simultaneously,
the |\vfill| effectively prevents |\vfil| from stretching. You can think
of it as if\/ |\vfil| has one mile of stretchability, while |\vfill| has
a trillion miles.

\danger Besides |\vfil| and |\vfill|, \TeX\ has ^|\hfil| and ^|\hfill|,
for stretching indefinitely in the horizontal direction. You can also say
^|\hss| or ^|\vss|, in order to get glue that is infinitely shrinkable as
well as infinitely stretchable.  \ (The name `|\hss|' stands for ``horizontal
stretch or shrink''; `|\vss|' is its vertical counterpart.) \ Finally, the
primitives ^|\hfilneg| and ^|\vfilneg| will cancel the stretchability of
|\hfil| and |\vfil|; we shall discuss applications of these curious glues later.

\danger Here are some examples of\/ |\hfil|, using the ^|\line| macro of
plain \TeX, which creates an hbox whose width is the current |\hsize|:
^^{flush left}
\begintt
\line{This text will be flush left.\hfil}
\line{\hfil This text will be flush right.}
\line{\hfil This text will be centered.\hfil}
\line{Some text flush left\hfil and some flush right.}
\line{Alpha\hfil centered between Alpha and Omega\hfil Omega}
\line{Five\hfil words\hfil equally\hfil spaced\hfil out.}
\endtt

\dangerexercise Describe the result of \xdef\linexno{\the\exno}%
\begintt
\line{\hfil\hfil What happens now?\hfil}
\line{\hfill\hfil and now?\hfil}
\endtt
\answer `What happens now?'\ is placed in a line of width |\hsize|, with
twice as much space at the left as at the right; `and now?'\ is put flush right
on the following line.

\ddangerexercise How do the following three macros behave differently?
\begintt
\def\centerlinea#1{\line{\hfil#1\hfil}}
\def\centerlineb#1{\line{\hfill#1\hfill}}
\def\centerlinec#1{\line{\hss#1\hss}}
\endtt
\answer The first two give an ``overfull box'' if the argument doesn't fit
on a line; the third allows the argument to stick out into the margins
instead. \ (Plain \TeX's ^|\centerline| is |\centerlinec|; the stickout effect
shows up in the narrow-column experiment of Chapter~6.) \ If the argument
contains no infinite glue, |\centerlinea| and |\centerlineb| produce the same
effect; but |\centerlineb| will center an argument that contains `fil' glue.

\danger In order to specify such infinities, you are allowed to use
the special units `^|fil|', `^|fill|', and `^|filll|' in the \<dimen>
parts of a stretchability or shrinkability component. For example,
|\vfil|, |\vfill|, |\vss|, and |\vfilneg| are essentially equivalent
to the glue specifications
\begintt
\vskip 0pt plus 1fil
\vskip 0pt plus 1fill
\vskip 0pt plus 1fil minus 1fil
\vskip 0pt plus -1fil
\endtt
respectively. It's usually best to stick to the first order infinity
(fil) as much as you can, resorting to second order (fill) only when
you really need something extremely infinite. Then the ultimate order (filll) is
always available as a last resort in emergencies. \ (\TeX\ does not
provide a `^|\vfilll|' primitive, since the use of this highest infinity
is not encouraged.) \ You can use fractional multiples of infinity like
`|3.25fil|', as long as you stick to fewer than 16384 fil units.  \TeX\
actually does its calculations with integer multiples of $\rm2^{-16}\,fil$
(or fill or filll); so |0.000007filll| turns out to be indistinguishable
from |0pt|, but |0.00001filll| is infinitely greater than |16383.99999fill|.

Now here's something important for all \TeX nical typists to know:
Plain \TeX\ puts extra space at the end of a ^{sentence}; furthermore,
it automatically increases the stretchability (and decreases
the shrinkability) after ^{punctuation} marks.  The reason is that it's
usually better to put more space after punctuation than between two ordinary
words, when spreading a line out to reach the desired margins. Consider, for
example, the following sentences from a classic kindergarten pre-primer:
^^{Dick and Jane}
\begintt
``Oh, oh!'' cried Baby Sally. Dick and Jane laughed.
\endtt
If \TeX\ sets this at its natural width, all the spaces will be the same,
except after the quote and after `Baby Sally.':
\begindisplay
``Oh, oh!'' cried Baby Sally. Dick and Jane laughed.
\enddisplay
But if the line needs to be expanded by 5 points, 10 points, 15 points, or more,
\TeX\ will set it as
\begindisplay
\hbox spread 5pt{``Oh, oh!'' cried Baby Sally. Dick and Jane laughed.}\cr
\hbox spread 10pt{``Oh, oh!'' cried Baby Sally. Dick and Jane laughed.}\cr
\hbox spread 15pt{``Oh, oh!'' cried Baby Sally. Dick and Jane laughed.}\cr
\hbox spread 20pt{``Oh, oh!'' cried Baby Sally. Dick and Jane laughed.}\cr
\enddisplay
The glue after the comma stretches at 1.25 times the rate of the
glue between adjacent words; the glue after the period and after the |!''|
stretches at 3 times the rate.  There is no glue between adjacent letters,
so individual words will always look the same.  If \TeX\ had to shrink
this line to its minimum width, the result would be
\begindisplay \hfuzz 1000pt
\hbox to 0pt{``Oh, oh!'' cried Baby Sally. Dick and Jane laughed.}
\enddisplay
The glue after a ^{comma} shrinks only 80 percent as much as ordinary
interword glue, and after a ^{period} or ^{exclamation point} or
^{question mark} it shrinks by only one third as much.

This all makes for nice-looking output, but it unfortunately adds a bit
of a burden to your job as a typist, because \TeX's rule for determining
the end of a sentence {\sl doesn't always work}. The problem is that
a period sometimes comes in the middle of a sentence $\ldots$ like when it
is used (as here) to make an ``^{ellipsis}'' of three dots.
^^{three dots, see ellipsis} ^^{dot dot dot, see ellipsis}

Moreover, if you try to specify `$\ldots$' by typing three periods in
a row, you get `...'---the dots are too close together. One way to
handle this is to go into {\sl mathematics\/} mode, using the ^|\ldots|
control sequence defined in plain \TeX\ format. For example, if you type
\begintt
Hmmm $\ldots$ I wonder why?
\endtt
the result is `Hmmm $\ldots$ I wonder why?'. This works because
math formulas are exempt from the normal text spacing rules.
Chapter 18 has more to say about |\ldots| and related topics.

^{Abbreviations} present problems too. For example, the short story in
Chapter~6 referred to `Mr.~^{Drofnats}'; \TeX\ must be told somehow that the
period after `Mr.'\ or `Mrs.'\ or `Ms.'\ or `Prof.'\ or `Dr.'\ or `Rt.~Hon.',
etc., doesn't count as a sentence-ending ^{full stop}.

We avoided that embarrassment in Chapter~6 by typing `|Mr.~Drofnats|';
the ``^{tie}'' mark |~| ^^{tilde} tells plain \TeX\ to insert a normal
space, and to refrain from breaking between lines at that space. Another way
to get \TeX\ to put out a normal space is to type `|\|\]' (^{control
space}); e.g., `|Mr.\ Drofnats|' would be almost the same as `|Mr.~Drofnats|',
except that a line might end after the `Mr.'.

The tie mark is best for abbreviations within a name, and after several
other common abbreviations like `Fig.'\ and `cf.'\ and `vs.'\ and `resp.';
you will find that it's easy to train yourself to type `|cf.~Fig.~5|'.
In fact, it's usually wise to type |~| (instead of a space) just after a common
abbreviation that occurs in the middle of a sentence.
Manuals of style will tell you that the abbreviations `e.g.'\ and `i.e.'\
should always be followed by commas, never by spaces, so those particular
cases shouldn't need any special treatment.

The only remaining abbreviations that arise with significant frequency
occur in bibliographic references; ^{control spaces} are appropriate here.
^^{interword spacing}
If, for example, you are typing a manuscript that refers to
`Proc.\ Amer.\ Math.\ Soc.', you should say
\begintt
Proc.\ Amer.\ Math.\ Soc.
\endtt
Granted that this input looks a bit ugly, it makes the output look right.
It's one of the things we occasionally must do when dealing with a computer
that tries to be smart.

\exercise Explain how to type the following sentence: ``Mr.~\& Mrs.~^{User}
were married by Rev.~^{Drofnats}, who preached on
Matt.~19\thinspace:\thinspace3--9.''
\answer |Mr.~\& Mrs.~User
were married by Rev.~Drofnats, who preached on
Matt.~19\thinspace:\thinspace3--9.| \ (Such thin spaces are traditional
for ^{Biblical references} to chapter and verse, but you weren't really
expected to know that. Plain \TeX\ defines ^|\thinspace| to be a kern,
not glue; hence no break between lines will occur at a thinspace.)

\exercise Put the following bibliographic reference into plain \TeX\
language: Donald~E. ^{Knuth}, ``Mathematical typography,'' {\sl Bull.\
Amer.\ Math.\ Soc.\ \bf1} (1979), 337--372.
\answer |Donald~E.\ Knuth, ``Mathematical typography,'' {\sl Bull.\
Amer.\ Math.\ Soc.\ \bf1} (1979), 337--372.| \ (But the `|\|' after `|E.|'
isn't necessary, because of a rule you will learn if you venture
around the next dangerous bend.)

On the other hand, if you don't care about such refinements of spacing
you can tell plain \TeX\ to make all spaces the same, regardless of
punctuation marks, by simply typing `^|\frenchspacing|' at the beginning
of your manuscript. French spacing looks like this:
\begindisplay \frenchspacing
``Oh, oh!'' cried Baby Sally. Dick and Jane laughed.
\enddisplay
You can also shift back and forth between the two styles, either by saying
`^|\nonfrenchspacing|' to establish ^{sophisticated spacing}, or by making
your use of\/ |\frenchspacing| local to some group. For example, you might
want to use French spacing only when typing the bibliography of some document.

\danger \TeX\ doesn't consider a period or question mark or exclamation point
to be the end of a sentence if the preceding character is an uppercase
letter, since \TeX\ assumes that such uppercase letters are most likely
somebody's initials.  Thus, for example, the `|\|' is unnecessary after
the~`|I.|' in `|Dr.~Livingstone~I.\ Presume|'; that particular period is
not assumed to be a full stop. ^^{Presume}

\dangerexercise What can you do to make \TeX\ recognize the ends of sentences
that do end with uppercase letters (e.g., `$\ldots$ launched by NASA.' or
`Did I?' or `$\ldots$ see Appendix~A.')?
\answer There are several ways; perhaps the easiest are to type
`|\hbox{NASA}.|'\ or `|NASA\null.|' \ (The ^|\null| macro is an abbreviation
for `|\hbox{}|'.)

\danger You can see the glue that \TeX\ puts between words by looking at
the contents of hboxes in the internal ^{diagnostic format} that we
discussed ^^{internal box-and-glue representation} briefly in Chapter~11.
For example, Baby Sally's exclamation begins as follows, after \TeX\ has
digested it and put it into a box, assuming |\nonfrenchspacing|:
\begintt
.\tenrm \ (ligature ``)
.\tenrm O
.\tenrm h
.\tenrm ,
.\glue 3.33333 plus 2.08331 minus 0.88889
.\tenrm o
.\tenrm h
.\tenrm !
.\tenrm " (ligature '')
.\glue 4.44444 plus 4.99997 minus 0.37036
.\tenrm c
.\tenrm r
.\tenrm i
.\tenrm e
.\tenrm d
.\glue 3.33333 plus 1.66666 minus 1.11111
.\tenrm B
.\tenrm a
.\tenrm b
.\kern-0.27779
.\tenrm y
.\glue 3.33333 plus 1.66666 minus 1.11111
.\tenrm S
.\tenrm a
.\tenrm l
.\tenrm l
.\tenrm y
.\kern-0.83334
.\tenrm .
.\glue 4.44444 plus 4.99997 minus 0.37036
\endtt
The normal ^{interword glue} in font |\tenrm| is $3.33333\pt$, plus
$1.66666\pt$ of stretchability, minus $1.11111\pt$ of shrinkability.
Notice that the interword |\glue| in this list stretches more, and shrinks
less, after the punctuation marks; and the natural space is in fact larger
at the end of each sentence. This example also shows several other things
that \TeX\ does while it processes the sample line of text: It converts
|``| and |''| into single characters, i.e., ^{ligatures}; and it inserts
small ^{kerns} in two places to improve the spacing. A ^|\kern| is similar
to glue, but it is not the same, because kerns cannot stretch or shrink;
furthermore, \TeX\ will never break a line at a kern, unless that kern is
immediately followed by glue.

\ddanger You may be wondering what \TeX's rules for interword glue really
are, exactly. For example, how did \TeX\ remember the effect of Baby Sally's
exclamation point, when quotation marks intervened before the next space?
The details are slightly tricky, but not incomprehensible. When \TeX\
is processing a horizontal list of boxes and glue, it keeps track of a positive
integer called the current ``^{space factor}.'' The space factor is normally
1000, which means that the interword glue should not be modified. If the
space factor $f$ is different from 1000, the interword glue is computed as
follows:  Take the normal space glue for the current font, and add the
extra space if $f\ge2000$. \ (Each font specifies a normal space, normal
stretch, normal shrink, and extra space; for example, these quantities are
$3.33333\pt$, $1.66666\pt$, $1.11111\pt$, and $1.11111\pt$, respectively, in
^|cmr10|.  We'll discuss such font parameters in greater detail later.) \
^^|\fontdimen|
Then the stretch component is multiplied by $f/1000$, while the shrink
component is multiplied by $1000/f$.

\ddanger However, \TeX\ has two parameters ^|\spaceskip| and ^|\xspaceskip|
that allow you to override the normal spacing of the current font. If
$f\ge2000$ and if\/ |\xspaceskip| is nonzero, the |\xspaceskip| glue is
used for an ^{interword space}. Otherwise if\/ |\spaceskip| is nonzero,
the |\spaceskip| glue is used, with stretch and shrink components
multiplied by $f/1000$ and $1000/f$. For example, the ^|\raggedright|
macro of plain \TeX\ uses |\spaceskip| and |\xspaceskip| to suppress all
stretching and shrinking of interword spaces.

\ddanger The space factor $f$ is 1000 at the beginning of a horizontal list,
and it is set to 1000 just after a non-character box or a math formula
has been put onto the current horizontal list. You can say
`^|\spacefactor||=|\<number>' to assign any particular value to the space
factor; but ordinarily, $f$ gets set to a number other than 1000 only when
a simple character box goes on the list. Each character has a ^{space factor
code}, and when a character whose space factor code is $g$ enters the current
list the normal procedure is simply to assign $g$ as the new space factor.
However, if $g$ is zero, $f$ is not changed; and if $f<1000<g$, the
space factor is set to~1000. \ (In other words, $f$ doesn't jump from a
value less than~1000 to a value greater than~1000 in a single step.) \
The maximum space factor is 32767 (which is much higher
than anybody would ever want to use).

\ddanger When ^|INITEX| creates a brand new \TeX, all characters have
a space factor code of~1000, except that the uppercase letters `|A|'
through~`|Z|' have code~999. \ (This slight difference is what makes
punctuation act differently after an uppercase letter; do you see why?)
\ Plain \TeX\ redefines a few of these codes using the ^|\sfcode|
primitive, which is similar to |\catcode| (see Appendix~B\null); for
example, the instructions
\begintt
\sfcode`)=0      \sfcode`.=3000
\endtt
make right parentheses ``transparent'' to the space factor, while tripling
the stretchability after periods. The |\frenchspacing| operation resets
|\sfcode`.| to 1000.

\ddanger When ligatures are formed, or when a special character is
specified via ^|\char|, the space factor code is computed from the
individual characters that generated the ligature. For example,
plain \TeX\ sets the space factor code for single-right-quote to zero, so
that the effects of punctuation will be propagated.  Two adjacent
characters |''| combine to form a ligature that is in character position
\oct{042}; but the space factor code of this double-right-quote ligature
is never examined by \TeX, so plain \TeX\ does not assign any value to
|\sfcode'042|.

\ddangerexercise What are the space factors after each token
of the Dick-and-Jane example?
\answer 1000, except: 999 after |O|, |B|, |S|, |D|, and |J|; 1250 after the
comma; 3000 after the exclamation point, the right-quote marks, and the
periods. If a period had come just after the |B| (i.e., if the text had
said `|B. Sally|'), the space factor after that period would have
been~1000, not~3000.

\danger Here's the way \TeX\ goes about ^{setting the glue} when an hbox
is being wrapped up: The natural width, $x$, of the box contents is
determined by adding up the widths of the boxes and kerns inside, together
with the natural widths of all the glue inside. Furthermore the total
amount of glue stretchability and shrinkability in the box is computed;
let's say that there's a total of $y_0+y_1\,{\rm fil}+y_2\,{\rm
fill}+y_3\,{\rm filll}$ available for stretching and $z_0+z_1\,{\rm
fil}+z_2\,{\rm fill}+z_3\,{\rm filll}$ available for shrinking. Now the
natural width~$x$ is compared to the desired width~$w$.  If $x=w$,
all glue gets its natural width.
Otherwise the glue will be modified, by computing a ``^{glue set
ratio}''~$r$ and a ``^{glue set order}''~$i$ in the following way: \
(a)~If $x<w$, \TeX\ attempts to stretch the contents of the box; the glue
order is the highest subscript~$i$ such that $y_i$ is nonzero, and the
glue ratio is $r=(w-x)/y_i$. (If $y_0=y_1=y_2=y_3=0$, there's no
stretchability; both $i$ and $r$ are set to zero.) \ (b)~If $x>w$, \TeX\
attempts to shrink the contents of the box in a similar way; the glue
order is the highest subscript~$i$ such that $z_i\ne0$, and the glue ratio
is normally $r=(x-w)/z_i$. However, $r$ is set to 1.0 in the case $i=0$
and $x-w>z_0$, because the maximum shrinkability must not be exceeded. \
(c)~Finally, every glob of glue in the horizontal list being boxed is
modified. Suppose the glue has natural width~$u$, stretchability~$y$, and
shrinkability~$z$, where $y$~is a $j$th order infinity and $z$~is a $k$th
order infinity. Then if $x<w$ (stretching), this glue takes the new width
$u+ry$ if $j=i$; it keeps its natural width~$u$ if $j\ne i$.  If $x>w$
(shrinking), this glue takes the new width $u-rz$ if $k=i$; it keeps
its natural width~$u$ if $k\ne i$.  Notice that stretching or shrinking
occurs only when the glue has the highest order of infinity that doesn't
cancel out.

\danger \TeX\ will construct an hbox that has a given width $w$ if you issue
the command `\hbox{|\hbox to |\<dimen>|{|\<contents of box>|}|}', where
$w$ is the value of the \<dimen>. For example, the ^|\line| macro discussed
earlier in this chapter is simply an abbreviation for `|\hbox to\hsize|'.
^^|to|^^|\hbox|
\TeX\ also allows you to specify the exact amount of stretching or shrinking;
the command `\hbox{|\hbox spread|\<dimen>|{|\<contents of box>|}|}'
creates a box whose width~$w$ is a given amount more ^^|spread|
than the natural width of the contents. For example, one of the boxes
displayed earlier in this chapter was generated by
\begintt
\hbox spread 5pt{``Oh, oh!'' ... laughed.}
\endtt
In the simplest case, when you just want a box to have its natural width,
you don't have to write `|\hbox spread 0pt|'; you can simply say
`|\hbox{|\<contents of box>|}|'.

\danger The ^{baseline} of a constructed hbox is the common baseline of the
boxes inside. \ (More precisely, it's the common baseline that they would
share if they weren't raised or lowered.) \ The height and depth of a
constructed hbox are determined by the maximum distances by which the
interior boxes reach above and below the baseline, respectively. The
result of\/ |\hbox| never has negative height or negative depth, but the
width can be negative.

\dangerexercise Assume that
 |\box1| is $1\pt$~high, $1\pt$~deep, and $1\pt$~wide;
 |\box2| is $2\pt$~high, $2\pt$~deep, and $2\pt$~wide.
A third box is formed by saying ^^|\setbox|
\begintt
\setbox3=\hbox to3pt{\hfil\lower3pt\box1\hskip-3pt plus3fil\box2}
\endtt
What are the height, depth, and width of\/ |\box3|? Describe the position
of the reference points of boxes 1 and~2 with respect to the reference
point of box~3.
\answer |\box3| is $2\pt$ high, $4\pt$ deep, $3\pt$ wide.
Starting at the reference point of\/ |\box3|, go right $.75\pt$ and down
$3\pt$ to reach the reference point of\/ |\box1|; or go right $1\pt$
to reach the reference point of\/ |\box2|.

\danger The process of setting glue for vboxes is similar to that for
hboxes; but before we study the |\vbox| operation, we need to discuss how
\TeX\ stacks boxes up vertically so that their baselines tend to be
a fixed distance apart. The boxes in a horizontal list often touch each
other, but it's usually wrong to do this in a vertical list; imagine
how awful a page would look if its lines of type were brought closer
together whenever they didn't contain tall letters, or whenever they
didn't contain any letters that descended below the baseline.

\danger \TeX's solution to this problem involves three primitives called
^|\baselineskip|, ^|\lineskip|, and ^|\lineskiplimit|. A format designer
chooses values of these three quantities by writing
\begindisplay
|\baselineskip=|\<glue>\cr
|\lineskip=|\<glue>\cr
|\lineskiplimit=|\<dimen>\cr
\enddisplay
and the interpretation is essentially this: Whenever a box is added to a
vertical list, \TeX\ inserts ``^{interline glue}'' intended to make
the distance between the baseline of the new box and the baseline of
the previous box exactly equal to the value of\/ |\baselineskip|.
But if the interline glue calculated by this rule would cause the
top edge of the new box to be closer than |\lineskiplimit| to the
bottom edge of the previous box, then |\lineskip| is used as the
interline glue. In other words, the distance between adjacent baselines
will be the |\baselineskip| setting, unless that would bring the boxes
too close together; the |\lineskip| glue will separate adjacent boxes
in the latter case.

\danger The rules for interline glue in the previous paragraph are
carried out without regard to other kinds of glue that might be present;
all vertical spacing due to explicit appearances of\/ |\vskip| and |\kern|
acts independently of the interline glue. Thus, for example, a
^|\smallskip| between two lines always makes their baselines further apart
than usual, by the amount of a |\smallskip|; it does not
affect the decision about whether |\lineskip| glue is used between
those lines.

\danger For example, let's suppose that |\baselineskip=12pt plus 2pt|,
|\lineskip=|\penalty0|3pt minus 1pt|, and |\lineskiplimit=2pt|. \ (These values
aren't particularly useful; they have simply been chosen to illustrate
the rules.) \ Suppose further that a box whose depth is $3\pt$
was most recently added to the current vertical list; we are about to
add a new box whose height is~$h$. If $h=5\pt$, the interline glue
will be $4\pt$~plus~$2\pt$, since this will make the baselines
$12\pt$~plus~$2\pt$ apart when we add $h$ and the previous depth to
the interline glue. But if $h=8\pt$, the interline
glue will be $3\pt$~minus~$1\pt$, since |\lineskip| will be
chosen in order to keep from violating the given |\lineskiplimit|
when stretching and shrinking are ignored.

\danger When you are typesetting a document that spans several pages,
it's generally best to define the |\baselineskip| so that it cannot stretch
or shrink, because this will give more uniformity to the pages. A small
variation in the distance between baselines---say only half a point---can
make a substantial difference in the appearance of the type, since it
significantly affects the proportion of white to black. On the other hand,
if you are preparing a one-page document, you might want to give the
baselineskip some stretchability, so that \TeX\ will help you fit the copy
on the page.

\dangerexercise What settings of\/ |\baselineskip|, |\lineskip|, and
|\lineskiplimit| will cause the interline glue to be a ``continuous''
function of the next box height (i.e., the interline glue will never
change a lot when the box height changes only a little)?
\answer The stretch and shrink components of\/ |\baselineskip| and
|\lineskip| should be equal, and the |\lineskiplimit| should
equal the normal |\lineskip| spacing, to guarantee continuity.

\danger A study of \TeX's ^{internal box-and-glue representation} should
help to firm up some of these ideas. Here is an excerpt from the vertical
list that \TeX\ constructed when it was typesetting this very paragraph:
\begintt
\glue 6.0 plus 2.0 minus 2.0
\glue(\parskip) 0.0 plus 1.0
\glue(\baselineskip) 1.25
\hbox(7.5+1.93748)x312.0, glue set 0.80154, shifted 36.0 []
\penalty 10000
\glue(\baselineskip) 2.81252
\hbox(6.25+1.93748)x312.0, glue set 0.5816, shifted 36.0 []
\penalty 50
\glue(\baselineskip) 2.81252
\hbox(6.25+1.75)x348.0, glue set 116.70227fil []
\penalty 10000
\glue(\abovedisplayskip) 6.0 plus 3.0 minus 1.0
\glue(\lineskip) 1.0
\hbox(149.25+0.74998)x348.0 []
\endtt
{\showboxdepth0\showboxbreadth9999\batchmode\showlists\errorstopmode}%
% The log file now contains lines like these; I copied them
% into this MS! But I deleted an `insert' for the index...
The first |\glue| in this example is the ^|\medskip| that precedes each
dangerous-bend paragraph. Then comes the ^|\parskip| glue, which is
automatically supplied before the first line of a new paragraph. Then
comes some interline glue of $1.25\pt$; it was calculated to make
a total of $11\pt$ when the height of the next box ($7.5\pt$)
and the depth of the previous box were added. \ (The previous box is not
shown---it's the bottom line of exercise~\chapno.\the\exno---but
we can deduce that its depth was $2.25\pt$.) \ The |\hbox| that
follows is the first line of this paragraph; it has been shifted right
$36\pt$ because of ^{hanging indentation}. The glue set ratio
for this hbox is 0.80154; i.e., the glue inside is stretched by 80.154\%\
of its stretchability. \ (In the case of shrinking, the ratio following
`^|glue set|' would have been preceded by `|- |'; hence we know that
stretching is involved here.) \ \TeX\ has put `|[]|' at the end of each
hbox line to indicate that there's something in the box that isn't shown.
\ (The box contents would have been displayed completely, if
^|\showboxdepth| had been set higher.) \ The ^|\penalty| indications are
used to discourage bad breaks between pages, as we will see later. The
third hbox has a glue ratio of 116.70227, which applies to
first-order-infinite stretching (i.e., fil); this results from an |\hfil|
that was implicitly inserted just before the displayed material, to fill
up the third line of the paragraph. Finally the big hbox whose height is
$149.25\pt$ causes |\lineskip| to be the interline glue. This large box
contains the individual lines of typewriter type that are displayed; they
have been packaged into a single box so that they cannot be split between
pages. Careful study of this example will teach you a lot about \TeX's
inner workings.

\danger Exception: No interline glue is inserted before or after a rule
box. You can also inhibit interline glue by saying ^|\nointerlineskip|
between boxes.

\ddanger \TeX's implementation of interline glue involves another primitive
quantity called ^|\prevdepth|, which usually contains the depth of the
most recent box on the current vertical list. However, |\prevdepth| is set
to the sentinel value $-1000\pt$ at the beginning of a vertical list, or
just after a rule box; this serves to suppress the next interline glue.
The user can change the value of\/ |\prevdepth| at any time when building
a vertical list; thus, for example, the |\nointerlineskip| macro of
Appendix~B simply expands to `|\prevdepth=-1000pt|'.

\ddanger Here are the exact rules by which \TeX\ calculates the interline
glue between boxes: Assume that a new box of height~$h$ (not a rule box)
is about to be appended to the bottom of the current vertical list, and
let $\hbox{|\prevdepth|}=p$, $\hbox{|\lineskiplimit|}=l$,
\hbox{$\hbox{|\baselineskip|}=(b$ plus~$y$ minus~$z)$}.  If $p\le-1000\pt$,
no interline glue is added.  Otherwise if $b-p-h\ge l$, the interline glue
`$(b-p-h)$ plus~$y$ minus~$z$' will be appended just above the new box.
Otherwise the |\lineskip| glue will be appended. Finally, |\prevdepth| is
set to the depth of the new box.

\ddangerexercise Mr.~B. L. ^{User} had an application in which he wanted
to put a number of boxes together in a vertical list, with no space
between them. He didn't want to say |\nointerlineskip| after each box;
so he decided to set |\baselineskip|, |\lineskip|, and |\lineskiplimit|
all equal to |0pt|. Did this work?
\answer Yes it did, but only because none of his boxes had a negative
height or depth. He would have been safer if he had set
|\baselineskip=-1000pt|, |\lineskip=0pt|, and
|\lineskiplimit=16383pt|. \ (Plain \TeX's ^|\offinterlineskip| macro does this.)

\danger The vertical analog of\/ |\hbox| is ^|\vbox|, and \TeX\ will obey
the commands `|\vbox to|\<dimen>' and `|\vbox spread|\<dimen>' in about
the way you would expect, by analogy with the horizontal case. However,
there's a slight complication because boxes have both height and depth
in the vertical direction, while they have only width in the
horizontal direction. The dimension in a |\vbox| command refers to
the final height of the vbox, so that, for example, `|\vbox to 50pt{...}|'
produces a box that is $50\pt$ high; this is appropriate because everything
that can stretch or shrink inside a vbox appears in the part that
contributes to the height, while the depth is unaffected by glue setting.

\danger The depth of a constructed |\vbox| is best thought of as the depth
^^{depth of box} ^^{height of box}
of the bottom box inside. Thus, a vbox is conceptually built by taking
a bunch of boxes and arranging them so that their reference
points are lined up vertically; then the reference point of the lowest
box is taken as the reference point of the whole, and the glue is set
so that the final height has some desired value.

\danger However, this description of vboxes glosses over some
technicalities that come up when you consider unusual cases. For example,
\TeX\ allows you to shift boxes in a vertical list to the right or to
the left by saying
`^|\moveright|\<dimen>\<box>' or `^|\moveleft|\<dimen>\<box>'; this is
like the ability to ^|\raise| or ^|\lower| boxes in a horizontal list, and
it implies that the reference points inside a vbox need not always lie in
a vertical line.  Furthermore, it is necessary to guard against boxes that
have too much depth, lest they extend too far into the bottom margin of a
page; and later chapters will point out that vertical lists can contain
other things like penalties and marks, in addition to boxes and glue.

\ddanger Therefore, the actual rules for the depth of a constructed vbox
are somewhat \TeX nical. Here they are: Given a vertical list that
is being wrapped up via |\vbox|, the problem is to determine its
natural depth. \ (1)~If the vertical list contains no boxes, the depth
is zero. \ (2)~If there's at least one box, but if the final box is
followed by kerning or glue, possibly with intervening penalties or
other things, the depth is zero. \ (3)~If there's at least
one box, and if the final box is not followed by kerning or glue, the
depth is the depth of that box. \ (4)~However, if the depth computed
by rules (1), (2), or~(3) exceeds ^|\boxmaxdepth|, the depth will be
the current value of\/ |\boxmaxdepth|. \ (Plain \TeX\ sets |\boxmaxdepth|
to the largest possible dimension; therefore rule~(4) won't apply unless
you specify a smaller value. When rule~(4) does decrease the depth,
\TeX\ adds the excess depth to the box's natural height, essentially
moving the reference point down until the depth has been reduced to the
stated maximum.)

\danger The glue is set in a vbox just as in an hbox, by determining a
^{glue set ratio} and a ^{glue set order}, based on the difference between
the natural height~$x$ and the desired height~$w$, and based on the amounts
of stretchability and shrinkability that happen to be present.

\danger The width of a computed |\vbox| is the maximum distance by which
an enclosed box extends to the right of the reference point, taking
possible shifting into account. This width is always nonnegative.

\dangerexercise Assume that |\box1| is $1\pt$ high, $1\pt$ deep, and
$1\pt$ wide; |\box2| is $2\pt$ high, $2\pt$ deep, and $2\pt$ wide; the
baselineskip, lineskip, and lineskiplimit are all zero; and the |\boxmaxdepth|
is very large. A third box is formed by saying
\begintt
\setbox3=\vbox to3pt{\moveright3pt\box1\vskip-3pt plus3fil\box2}
\endtt
What are the height, depth, and width of\/ |\box3|? Describe the position
of the reference points of boxes 1 and~2 with respect to the reference point
of box~3.
\answer The interline glue will be zero, and the natural height is
$1+1-3+2=1\pt$ (because the depth of\/ |\box2| isn't included in the natural
height); so the glue will ultimately become |\vskip-1pt| when it's set.
Thus, |\box3| is $3\pt$ high, $2\pt$ deep, $4\pt$ wide. Its reference
point coincides with that of\/ |\box2|; to get to the reference point
of\/ |\box1| you go up $2\pt$ and right $3\pt$.

\ddangerexercise Under the assumptions of the previous exercise, but
with |\baselineskip=9pt minus3fil|, describe |\box4| after
\begintt
\setbox4=\vbox to4pt{\vss\box1\moveleft4pt\box2\vss}
\endtt
\answer The interline glue will be $6\pt$ minus $3\,{\rm fil}$; the final
depth will be zero, since |\box2| is followed by glue; the natural
height is $12\pt$; and the shrinkability is $5\,{\rm fil}$. So |\box4|
will be $4\pt$ high, $0\pt$ deep, $1\pt$ wide, and it will contain
five items: |\vskip|\penalty0\hbox{|-1.6pt|}, |\box1|, |\vskip1.2pt|,
|\moveleft4pt\box2|, |\vskip-1.6pt|. Starting at the reference point of
|\box4|, you get to the reference point of\/ |\box1| by going up $4.6\pt$,
or to the reference point of\/ |\box2| by going up $.4\pt$ and left $4\pt$.
\ (For example, you go up $4\pt$ to get to the upper left corner of
|\box4|; then down $-1.6\pt$, i.e., up $1.6\pt$, to get to the upper left
corner of\/ |\box1|; then down $1\pt$ to reach its reference point.  This
problem is clearly academic, since it's rather ridiculous to include
infinite shrinkability in the baselineskip.)

\ddangerexercise Solve the previous problem but with |\boxmaxdepth=-4pt|.
\answer Now |\box4| will be $4\pt$ high, $-4\pt$ deep, $1\pt$ wide, and it
will contain |\vskip|\penalty0\hbox{|-2.4pt|}, |\box1|, |\vskip-1.2pt|,
|\moveleft4pt\box2|, |\vskip-2.4pt|. From the baseline of\/ |\box4|, go up
exactly $5.4\pt$ to reach the baseline of\/ |\box1|, or exactly $3.6\pt$
to reach the baseline of\/ |\box2|.

\danger We have observed that |\vbox| combines a bunch of boxes into
a larger box that has the same baseline as the bottom box inside.
\TeX\ has another operation called ^|\vtop|, which gives you a box like |\vbox|
but with the same baseline as the top box inside. For example,
\begintt
\hbox{Here are \vtop{\hbox{two lines}\hbox{of text.}}}
\endtt
produces
\begindisplay
Here are \vtop{\hbox{two lines}\hbox{of text.}}
\enddisplay

\ddanger You can say `|\vtop to|\<dimen>' and `|\vtop spread|\<dimen>'
just as with |\vbox|, but you should realize what such a construction
means. \TeX\ implements |\vtop| as follows: \ (1)~First a~vertical box
is formed as if\/ |\vtop| had been |\vbox|, using all of the rules for
|\vbox| as given above. \ (2)~The final height~$x$ is defined to
be zero unless the very first item inside the new vbox is a box;
in the latter case, $x$ is the height of that box. \ (3)~Let $h$
and~$d$ be the height and depth of the vbox in step~(1). \TeX\ completes
the |\vtop| by moving the reference point up or down, if necessary, so
that the box has height~$x$ and depth $h+d-x$.

\ddangerexercise Describe the empty boxes that you get from
`|\vbox to|\<dimen>|{}|' and
`|\vtop to|\<dimen>|{}|'. What are their heights, depths, and widths?
\answer |\vbox to| $x$|{}| produces height $x$;
|\vtop to| $x$|{}| produces depth $x$; the other dimensions are zero.
\ (This holds even when $x$ is negative.)

\ddangerexercise Define a macro |\nullbox#1#2#3| that produces a box
whose height, depth, and width are given by the three parameters.
The box should contain nothing that will show up in print.
\answer There are several possibilities:
\begintt
\def\nullbox#1#2#3{\vbox to#1{\vss\hrule height-#2depth#2width#3}}
\endtt
works because the rule will be of zero thickness. Less tricky is
\begintt
\def\nullbox#1#2#3{\vbox to#1{\vss\vtop to#2{\vss\hbox to#3{}}}}
\endtt
Both of these are valid with negative height and/or depth, but they do
not produce negative width. If the width might be negative, but not the
height or depth, you can use, e.g.,
|\def\nullbox#1#2#3{\hbox to#3{\hss\raise#1\null\lower#2\null}}|.
It's impossible for |\hbox| to construct a box
whose height or depth is negative; it's impossible for |\vbox| or
|\vtop| to construct a box whose width is negative.\par
However, there's actually a trivial solution to the general problem, based on
features that will be discussed later:
\begintt
\def\nullbox#1#2#3{\setbox0=\null
  \ht0=#1 \dp0=#2 \wd0=#3 \box0 }
\endtt

\danger The |\vbox| operation tends to produce boxes with large height and
small depth, while |\vtop| tends to produce small height and large depth.
If you're trying to make a vertical list out of big vboxes, however, you
may not be satisfied with either |\vbox| or |\vtop|; you might well wish
that a box had two reference points simultaneously, one for the top and
one for the bottom. If such a dual-reference-point scheme were in use, one
could define interline glue based on the distance between the lower
reference point of one box and the upper reference point of its successor
in a vertical list. But alas, \TeX\ gives you only one reference point per box.

\danger There's a way out of this dilemma, using an important idea
called a ``^{strut}.'' Plain \TeX\ defines ^|\strut| to be an invisible
box of width zero that extends just enough above and below the baseline
so that you would need no interline glue at all if every line contained
a strut. \ (Baselines are $12\pt$ apart in plain \TeX; it turns out that
|\strut| is a vertical rule, $8.5\pt$ high and $3.5\pt$ deep and
$0\pt$ wide.) \ If you contrive to put a strut on the top line and
another on the bottom line, inside your large vboxes, then it's possible to
obtain the correct spacing in a larger assembly by simply letting the
boxes butt together. For example, the ^|\footnote| macro in Appendix~B
puts struts at the beginning and end of every footnote, so that the spacing
will be right when several footnotes occur together at the bottom
of some page. ^^{fitting boxes together}

\danger If you understand boxes and glue, you're ready to learn
the ^|\rlap| and ^|\llap| macros of plain \TeX; these names are abbreviations
for ``right ^{overlap}'' and ``left overlap.'' Saying `|\rlap{|\<something>|}|'
is like typesetting \<something> and then backing up as if you hadn't
typeset anything. More precisely, `|\rlap{|\<something>|}|' creates a box
of width zero, with `\<something>' appearing just at the right of that box
(but not taking up any space). The |\llap| macro is similar, but it does
the ^{backspacing} first; in other words, `|\llap{|\<something>|}|'
creates a box of width zero, with `\<something>' extending just to the
left of that box. Using typewriter type, for example, you can typeset
`{\tt\rlap/=}' by saying either `|\rlap/=|' or `|/\llap=|'. It's possible
to put text into the left margin using |\llap|, or into the right margin
using |\rlap|, because \TeX\ does not insist that the contents of a box
must be strictly confined within that box's boundaries. ^^{marginal notes}

\danger The interesting thing about |\rlap| and |\llap| is that they can be
done so simply with infinite glue. One way to define |\rlap| would be
\begintt
\def\rlap#1{{\setbox0=\hbox{#1}\copy0\kern-\wd0}}
\endtt
but there's no need to do such a lengthy computation. The actual definition
in Appendix~B is much more elegant, namely,
\begintt
\def\rlap#1{\hbox to 0pt{#1\hss}}
\endtt
and it's worth pondering why this works. Suppose, for example, that you're
doing |\rlap{g}| where the letter `g' is $5\pt$ wide. Since |\rlap| makes
an hbox of width $0\pt$, the glue represented by ^|\hss| must shrink by
$5\pt$.  Well, that glue has $0\pt$ as its natural width, but it has
infinite shrinkability, so it can easily shrink to $-5\pt$; and
`|\hskip-5pt|' is exactly what |\rlap| wants in this case.

\dangerexercise Guess the definition of\/ |\llap|, without peeking at
Appendices A or~B.
\answer |\def\llap#1{\hbox to 0pt{\hss#1}}|

\dangerexercise (This is a sequel to exercise 12.\linexno,
but it's trickier.) \ Describe the result of
\begintt
\line{\hfil A puzzle.\hfilneg}
\endtt
\answer You get `A' at the extreme left and `puzzle.\null' at the extreme right,
because the space between words has the only stretchability that is finite;
the infinite stretchability cancels out. \ (In this case, \TeX's rule
about ^{infinite glue} differs from what you would get in the limit if the
value of $1\,{\rm fil}$ were finite but getting larger and larger.
The true limiting behavior would stretch the text `A~puzzle.\null' in the
same way, but it would also move that text infinitely far away beyond the right
edge of the page.)

\endchapter

There was things which he stretched,
but mainly he told the truth.
\author MARK ^{TWAIN}{^^{Clemens}}, {\sl Huckleberry Finn\/} (1884) % chap1

\bigskip

Every shape exists only because of the space around it.
% Jeder Formwert lebt nur dank seiner Umgebung.
$\ldots$ Hence there is a `right' position for every shape in every situation.
% ... Daraus geht hervor, da\ss\ ein bestimmter Formwert in einer bestimmten
% Situation einen bestimmten Platz verlangt, der ihm gem\"a\ss\ ist.
If we succeed in finding that position, we have done our job.
% Gelingt es uns, diese Pl\"atze zu finden,
% so nennen wir die Arbeit vollendet.
\author JAN ^{TSCHICHOLD}, {\sl Typographische Gestaltung\/} (1935) % p64
% I've borrowed from the free translation in {\sl Asymmetric Typography}

\eject
\beginchapter Chapter 13. Modes

Just as people get into different moods, \TeX\ gets into different ``modes.'' \
(Except that \TeX\ is more predictable than people.) \ There are six ^{modes}:

\medskip
\item\bull^{Vertical mode}. [Building the main vertical list, from which
the pages of output are derived.]

\smallskip\item\bull
^{Internal vertical mode}. [Building a vertical list for a vbox.]

\smallskip\item\bull
^{Horizontal mode}. [Building a horizontal list for a paragraph.]

\smallskip\item\bull
^{Restricted horizontal mode}. [Building a horizontal list for an hbox.]

\smallskip\item\bull
^{Math mode}. [Building a mathematical formula to be placed in
a horizontal list.]

\smallskip\item\bull
^{Display math mode}. [Building a mathematical formula to be placed
on a line by itself, temporarily interrupting the current paragraph.]

\medskip\noindent In simple situations, you don't need to be aware of what
mode \TeX\ is in, because the computer just does the right thing.  But
when you get an error message that says `\thinspace|!|~|You| |can't| |do|
|such-and-such| |in| |restricted| |horizontal| |mode|\thinspace', a
knowledge of modes helps to explain why \TeX\ thinks you goofed.

Basically \TeX\ is in one of the vertical modes when it is preparing a list of
boxes and glue that will be placed vertically above and below one another on
the page; it's in one of the horizontal modes when it is preparing a list
of boxes and glue that will be strung out horizontally next to each other
with baselines aligned; and it's in one of the math modes when it is
reading a formula.

A play-by-play account of a typical \TeX\ job should make the mode idea clear:
At the beginning, \TeX\ is in vertical mode, ready to construct pages. If you
specify glue or a box when \TeX\ is in vertical mode, the glue or the box gets
placed on the current page below what has already been specified. For example,
the ^|\vskip| instructions in the sample run we discussed in Chapter~6
contributed vertical glue to the page; and the ^|\hrule| instructions
contributed horizontal rules at the top and bottom of the story. The
^|\centerline| commands also produced boxes that were included in the main
vertical list; but those boxes required a bit more work than the rule boxes:
\TeX\ was in vertical mode when it encountered
`|\centerline{\bf A SHORT STORY}|', and it went temporarily into restricted
horizontal mode while processing the words `|A SHORT STORY|'; then
the digestive process returned to vertical mode, after setting the
glue in the~|\centerline|~box.

Continuing with the example of Chapter 6, \TeX\ switched into horizontal
mode as soon as it read the `|O|' of `|Once upon a time|'. Horizontal mode
is the mode for making ^{paragraphs}. The entire paragraph (lines 7 to~11
of the |story| file) was input in horizontal mode; then the text was
divided into output lines of the appropriate width, those lines were
put in boxes and appended to the page (with appropriate interline glue between
them), and \TeX\ was back in vertical mode. The `|M|' on line~12 started up
horizontal mode again.

When \TeX\ is in vertical mode or internal vertical mode,
the first token of a new paragraph changes the mode to horizontal for the
duration of a paragraph. In other words, things that do not have a vertical
orientation cause the mode to switch automatically from vertical to
horizontal. This occurs when you type any character, or ^|\char| or ^|\accent|
or ^|\hskip| or |\|\] ^^{control space} or ^|\vrule| or math shift (|$|);
\TeX\ inserts the current paragraph ^{indentation} and rereads the
horizontal token as if it had occurred in horizontal mode.

\danger You can also tell \TeX\ explicitly to go into horizontal mode,
instead of relying on such implicit mode-switching, by saying `^|\indent|'
or `^|\noindent|'. For example, if line~7 of the |story| file in Chapter~6
had begun
\begintt
\indent Once upon a time, ...
\endtt
the same output would have been obtained, because `|\indent|' would have
instructed \TeX\ to begin the paragraph. And if that line had begun with
\begintt
\noindent Once upon a time, ...
\endtt
the first paragraph of the story would not have been indented. The
|\noindent| command simply tells \TeX\ to enter horizontal mode if the
current mode is vertical or internal vertical; |\indent| is similar,
but it also creates an empty box whose width is the current value of
^|\parindent|, and it puts this empty box into the current horizontal list.
Plain \TeX\ sets |\parindent=20pt|.
If you say |\indent\indent|, you get double indentation; if you say
|\noindent\noindent|, the second |\noindent| does nothing.

\dangerexercise If you say `^|\hbox||{...}|' in horizontal mode, \TeX\ will
construct the specified box and it will contribute the result to the
current paragraph. Similarly, if you say `|\hbox{...}|' in vertical mode,
\TeX\ will construct a box and contribute it to the current page.
What can you do if you want to begin a paragraph with an |\hbox|?
\answer Simply saying |\hbox{...}| won't work, since that box will just
continue the previous vertical list without switching modes. You need
to start the paragraph explicitly, and the straightforward way to
do that is to say |\indent\hbox{...}|.
But suppose you want to define
a macro that expands to an hbox, where this macro is to be used in the
midst of a paragraph as well as at the beginning; then you don't want
to force users to type |\indent| before calling your macro at the
beginning of a paragraph, nor do you want to say |\indent| in the
macro itself (since that might insert unwanted indentations).  One
solution to this more general problem is to say
`|\|\]^|\unskip||\hbox{...}|', since |\|\] makes the mode
horizontal while |\unskip| removes the unwanted space. Plain \TeX\
provides a ^|\leavevmode| macro, which solves this problem in what is
probably the most efficient way: |\leavevmode| is an abbreviation for
`|\unhbox\voidbox|', where |\voidbox| is a permanently empty box register.

When handling simple manuscripts, \TeX\ spends almost all of its time in
horizontal mode (making paragraphs), with brief excursions into vertical
mode (between paragraphs). A paragraph is completed when you type ^|\par|
or when your manuscript has a blank line, since a blank line is converted
to |\par| by the reading rules of Chapter~8. A paragraph also ends when
you type certain things that are incompatible with horizontal mode.
For example, the command `|\vskip 1in|' on line~16 of Chapter~6's |story|
file was enough to terminate the paragraph about `|...beautiful
documents.|'; no |\par| was necessary, since |\vskip| introduced
vertical glue that couldn't belong to the~paragraph.

If a begin-math token (|$|) appears in
horizontal mode, \TeX\ plunges into math mode and processes the formula up until
the closing `|$|', then appends the text of this formula to the current
paragraph and returns to horizontal mode. Thus, in the ``I wonder why?''\
example of Chapter~12, \TeX\ went into math mode temporarily
while processing |\ldots|, treating the dots as a formula.

However, if two consecutive begin-math tokens appear in a paragraph (|$$|),
\TeX\ interrupts the paragraph where it is, contributes the paragraph-so-far
to the enclosing vertical list, then processes a math formula in display
math mode, then contributes this formula to the enclosing list, then
returns to horizontal mode for more of the paragraph. \ (The formula to be
displayed should end with `|$$|'.) \ For example, suppose you type
\begintt
the number $$\pi \approx 3.1415926536$$ is important.
\endtt
\TeX\ goes into display math mode between the |$$|'s, and the output you
get states that the number $$\pi \approx 3.1415926536$$ is important.
^^|\pi|

\smallskip
\TeX\ ignores blank spaces and blank lines (or ^|\par| commands) when it's
in vertical or internal vertical mode, so you need not worry that
such things might change the mode or affect a printed document.
A ^{control space} (|\|\]) will, however, be regarded as the
beginning of a paragraph; the paragraph will start with a blank space
after the indentation.

\smallskip
At the end of a \TeX\ manuscript it's usually best to finish everything off
by typing `^|\bye|', which is plain \TeX's abbreviation for
`|\vfill\eject\end|'. The `|\vfill|' gets \TeX\ into vertical
mode and inserts enough space to fill up the last page; `|\eject|' outputs
that last page; and `|\end|' sends the computer into its ^{endgame} routine.

\danger \TeX\ gets into internal vertical mode when you ask it to construct
something from a vertical list of boxes (using |\vbox| or |\vtop| or |\vcenter|
or |\valign| or |\vadjust| or |\insert|). It gets into restricted
horizontal mode when you ask it to construct something from a horizontal list
of boxes (using |\hbox| or |\halign|). Box construction is discussed
in Chapters 12 and~21. We will see later that there is very little difference
between internal vertical mode and ordinary vertical mode, and very little
difference between restricted horizontal mode and ordinary horizontal mode;
but they aren't quite identical, because they have different goals.

\danger Whenever \TeX\ looks at a token of input to decide what should be
done next, the current mode has a potential influence on what that token
means. For example, ^|\kern| specifies vertical spacing in vertical mode,
but it specifies horizontal spacing in horizontal mode; a math shift
character like `|$|' causes entry to math mode from horizontal mode, but
it causes exit from math mode when it occurs in math mode; two consecutive
math shifts (|$$|) appearing in horizontal mode will initiate display
math mode, but in restricted horizontal mode they simply denote an empty
math formula.  \TeX\ uses the fact that some operations are inappropriate
in certain modes to help you recover from errors that might have crept
into your manuscript.  Chapters 24 to~26 explain exactly what
happens to every possible token in every possible mode.

\danger \TeX\ often interrupts its work in one mode to do some task in
another mode, after which the original mode is resumed again.
For example, you can say `|\hbox{|' in any mode; when \TeX\ digests this,
it suspends whatever else it was doing and enters restricted horizontal
mode. The matching `|}|' will eventually cause the hbox to be completed,
whereupon the postponed task will be taken up anew. In this sense \TeX\ can
be in many modes simultaneously, but only the innermost
mode influences the calculations at any time; the other modes have
been pushed out of \TeX's consciousness.

\goodbreak
\danger One way to become familiar with \TeX's modes is to consider the
following curious test file called |modes.tex|, which exercises all the
modes at once:
$$\halign{\hbox to\parindent{\hfil\sevenrm#\ \ }&#\hfil\cr
1&|\tracingcommands=1|\cr\noalign{^^|\tracingcommands|}
2&|\hbox{|\cr
3&|$|\cr
4&|\vbox{|\cr
5&|\noindent$$|\cr
6&|x\showlists|\cr
7&|$$}$}\bye|\cr}$$
The first line of ^|modes.tex| tells \TeX\ to log every command it
receives; \TeX\ will produce diagnostic data whenever
|\tracingcommands| is positive. Indeed, if you run \TeX\ on |modes.tex|
you will get a |modes.log| file that includes the following information:
\begintt
{vertical mode: \hbox}
{restricted horizontal mode: blank space  }
{math shift character $}
{math mode: blank space  }
{\vbox}
{internal vertical mode: blank space  }
{\noindent}
{horizontal mode: math shift character $}
{display math mode: blank space  }
{the letter x}
\endtt
The meaning is that \TeX\ first saw an |\hbox| token in vertical mode;
this caused it to go ahead and read the `|{|' behind the scenes.
Then \TeX\ entered restricted horizontal mode, and
saw the blank space token that resulted from the end of line~2 in the
file. Then it saw a math shift character token (still in restricted
horizontal mode), which caused a shift to math mode; another
blank space came through. Then |\vbox| inaugurated internal
vertical mode, and |\noindent| instituted horizontal mode within that; two
subsequent |$| signs led to display math mode. \ (Only the first |$| was
shown by |\tracingcommands|, because that one caused \TeX\ to look ahead
for another.)

\danger The next thing in |modes.log| after the output above is
`|{|^|\showlists||}|'. This is another handy diagnostic command that
you can use to find out things that \TeX\ ordinarily keeps to itself; it
causes \TeX\ to display the lists that are being worked on, in the current
mode and in all enclosing modes where the work has been suspended:
\begintt
### display math mode entered at line 5
\mathord
.\fam1 x
### internal vertical mode entered at line 4
prevdepth ignored
### math mode entered at line 3
### restricted horizontal mode entered at line 2
\glue 3.33333 plus 1.66666 minus 1.11111
spacefactor 1000
### vertical mode entered at line 0
prevdepth ignored
\endtt
In this case the lists represent five levels of activity, all
present at the end of line~6 of |modes.tex|.
The current mode is shown first, namely, display math mode, which began
on line~5. The current math list contains one ``^{mathord}''
object, consisting of the letter~|x| in family~1. \ (Have patience and you
will understand what that means, when you learn about \TeX's math formulas.) \
Outside of display math mode comes internal vertical mode, to which \TeX\
will return when the paragraph containing the
displayed formula is complete. The vertical list ^^|prevdepth ignored|
on that level is empty; `|prevdepth ignored|' means that |\prevdepth|
has a value $\le-1000\pt$, so that the next interline glue will be omitted
(cf.~Chapter~12). The math mode outside of this internal vertical mode has
an empty list, likewise, but the restricted horizontal mode enclosing the math
mode contains some glue. Finally, we see the main vertical mode that
encloses everything; this mode was `|entered at line 0|', i.e., before
the file |modes.tex| was input; nothing has been contributed
so far to the vertical list on this outermost level.

\dangerexercise Why is there glue in one of these lists but not in the
others?
\answer The output of\/ |\tracingcommands| shows that four blank space tokens
were digested; these originated at the ends of lines 2,~3, 4, and~5. Only
the first had any effect, since blank spaces are ignored in math formulas
and in vertical modes.

\dangerexercise After this output of\/ |\showlists|, the |modes.log| file
contains further output from |\tracingcommands|. In fact, the next two
lines of that file are
\begintt
{math shift character $}
{horizontal mode: end-group character }}
\endtt
because the `|$$|' on line 7 finishes the displayed formula, and this
resumes horizontal mode for the paragraph that was interrupted.
What do you think are the next three lines of |modes.log|\thinspace?
\answer The |end-group character| finishes the paragraph and the |\vbox|,
and |\bye| stands for `|\par\vfill...|', so the next three commands are
\begintt
{math mode: math shift character $}
{restricted horizontal mode: end-group character }}
{vertical mode: \par}
\endtt

\dangerexercise Suppose \TeX\ has generated a document without ever
leaving vertical mode. What can you say about that document?
\answer It contains only mixtures of vertical glue and horizontal rules
whose reference points appear at the left of the page; there's no text.

\ddangerexercise Some of \TeX's modes cannot immediately enclose other modes;
for example, display math mode is never directly enclosed by horizontal
mode, even though displays occur within paragraphs, because an interrupted
paragraph-so-far of horizontal mode is always completed and
removed from \TeX's memory before the processing of a displayed formula
begins. Give a complete characterization of all pairs of consecutive
modes that can occur in the output of\/ |\showlists|.
\answer Vertical mode can occur only as the outermost mode; horizontal
mode and display math mode can occur only when immediately enclosed by
vertical or internal vertical mode; ordinary math mode cannot be
immediately enclosed by vertical or internal vertical mode; all other
cases are possible.

\endchapter

Every mode of life has its conveniences.
\author SAMUEL ^{JOHNSON}, {\sl The Idler\/} (1758)

\bigskip

[Hindu musicians] have eighty-four modes,
of which thirty-six are in general use,
and each of which, it appears, has a peculiar expression,
and the power of moving some particular sentiment or affection.
\author MOUNTSTUART ^{ELPHINSTONE}, {\sl History of India\/} (1841)
% III.vii.I.297

\eject
\beginchapter Chapter 14. How \TeX\ Breaks\\Paragraphs into\\Lines

One of a typesetting system's chief duties is to take a long sequence of words
and to break it up into individual lines of the appropriate size.
For example, every paragraph of this manual has been broken into lines
that are 29~picas wide, but the author didn't have to worry about such
details when he composed the manuscript. \TeX\ chooses breakpoints
in an interesting way that considers each paragraph in its entirety; the
closing words of a paragraph can actually influence the appearance of the
first line. As a result, the spacing between words is as uniform as
possible, and the computer is able to reduce the number of times that
words must be hyphenated or formulas must be split between lines.
^^{H\&J, see hyphenation, line breaking, setting glue}
^^{justification, see setting glue, line breaking}
^^{quad left, see flush left}
^^{quad right, see flush right}
^^{quad middle, see :break}

The experiments of Chapter 6 have already illustrated the general ideas:
We discussed the notion of ``badness,'' and we ran into ``overfull'' and
``underfull'' boxes in difficult situations. We also observed that different
settings of \TeX's ^|\tolerance| parameter
will produce different effects; a higher tolerance
means that wider spaces are acceptable.

\TeX\ will find the absolutely best way to typeset any given paragraph,
according to its ideas of minimum badness. But such ``badness'' doesn't
account for everything, and if you rely entirely on
an automatic scheme you will occasionally encounter line breaks that are
not really the best on psychological grounds; this is inevitable, because
computers don't understand things the way people~do (at least not yet).
Therefore you'll sometimes want to tell the machine that certain places
are not good breakpoints. Conversely, you will sometimes want to force a
break at a particular spot. \TeX\ provides a convenient way to avoid
psychologically bad breaks, so that you will be able to obtain results of
the finest quality by simply giving a few hints to the machine.

``^{Ties}''---denoted by `|~|' in plain \TeX---are the key to
successful line breaking. ^^{auxiliary space, see tie} ^^{tilde}
Once you learn how to insert them, you will have graduated from the ranks
of ordinary \TeX nical typists to the select group of Distinguished
\TeX nicians. And it's really not difficult to train yourself to
insert occasional ties, almost without thinking, as you type a manuscript.
^^{line breaks, avoiding} ^^{breaks, avoiding bad}

When you type |~| it's the same as typing a space, except that \TeX\
won't break a line at this space. Furthermore, you shouldn't leave
any blanks next to the |~|, since they will count as additional spaces.
If you put |~| at the very end of a line in your input file, you'll get a
wider space than you want, because the \<return> that follows the |~|
produces an extra space.

We have already observed in Chapter~12 that it's generally a good idea to
type |~| after an abbreviation that does not come at the end of a sentence.
Ties also belong in several other places:

\smallskip
\item\bull In references to named parts of a document:
$$\halign{#\hfil&\hskip 80pt#\hfil\cr
|Chapter~12|&|Theorem~1.2|\cr
|Appendix~A|&|Table~\hbox{B-8}|\cr
|Figure~3|&|Lemmas 5 and~6|\cr}$$

\noindent(No |~| appears after `|Lemmas|' in the final example, since there's
no harm in having `5~and~6' at the beginning of a line. The use of\/ |\hbox|
is explained below.)

\smallbreak
\item\bull Between a person's forenames and between multiple surnames:
$$\halign{#\hfil&\hskip 40pt#\hfil\cr
|Donald~E. Knuth|&|Luis~I. Trabb~Pardo|\cr
|Bartel~Leendert van~der~Waerden|&|Charles~XII|\cr}$$

^^{Knuth} ^^{Trabb Pardo} ^^{van der Waerden} ^^{Charles XII}
\noindent
Note that it is sometimes better to hyphenate a name than to break it
between words; e.g., `Don-' and `ald~E.~Knuth' is more tolerable
than `Donald' and `E.~Knuth'. The previous rule can be regarded as a
special case of this one, since we may think of `Chapter~12' as a
compound name; another example is `|register~X|'. Sometimes a name is
so long that we dare not tie it all together, lest there be no way to
break the line: ^^{Vall\'ee Poussin}
\begintt
Charles Louis Xavier~Joseph de~la Vall\'ee~Poussin.
\endtt

\item\bull Between math symbols in apposition with nouns:
$$\halign{#\hfil\cr
|dimension~$d$       width~$w$       function~$f(x)$|\cr
|string~$s$ of length~$l$|\cr}$$

\noindent However, the last example should be compared with
\begintt
string~$s$ of length $l$~or more.
\endtt

\item\bull Between symbols in series:
$$\halign{#\hfil\cr
|1,~2, or~3|\cr
|$a$,~$b$, and~$c$.|\cr
|1,~2, \dots,~$n$.|\cr}$$

\item\bull When a symbol is a tightly bound object of a preposition:
$$\halign{#\hfil\cr
|of~$x$|\cr
|from 0 to~1|\cr
|increase $z$ by~1|\cr
|in common with~$m$.|\cr}$$

\noindent The rule does not, however, apply to compound objects:
\begintt
of $u$~and~$v$.
\endtt

\item\bull When mathematical phrases are rendered in words:
$$\halign{#\hfil&\hskip20pt#\hfil&\hskip20pt#\hfil\cr
|equals~$n$|&|less than~$\epsilon$|&|(given~$X$)|\cr
|mod~2|&|modulo~$p^e$|&|for all large~$n$|\cr}$$

\noindent Compare `|is~15|' with `|is 15~times the height|'.

\medbreak
\item\bull When cases are being enumerated within a paragraph:
^^{enumerated cases within a paragraph}
$$\halign{#\hfil\cr
|(b)~Show that $f(x)$ is (1)~continuous; (2)~bounded.|\cr}$$

\noindent It would be nice to boil all of these rules down to one or two
simple principles, and~it would be even nicer if the rules could be
automated so that keyboarding could be done without them; but subtle
semantic considerations seem to be involved. Therefore it's best to use
your own judgment with respect to ties. The computer needs your help.

A tie keeps \TeX\ from breaking at a space, but sometimes you want to
prevent the machine from breaking at a ^{hyphen} or a ^{dash}. This can be
done by using ^|\hbox|, because \TeX\ will not split up the contents of a
box; boxes are indecomposable units, once they have been constructed.  We
have already illustrated this principle in the `|Table~\hbox{B-8}|'
example considered earlier.  Another example occurs when you are typing
the page numbers in a ^{bibliographic reference}: It doesn't look good to
put \hbox{`22.'} on a line by itself, so you can type `|\hbox{13--22}.|'
to prohibit breaking `\hbox{13--22}.' On the other hand, \TeX\ doesn't
often choose line breaks at hyphens, so you needn't bother to insert
|\hbox| commands unless you need to correct a bad break that \TeX\ has
already made on a previous run.

\exercise Here are some phrases culled from previous chapters of this
manual. How do you think the author typed them?
\begindisplay
(cf.~Chapter~12).\cr
Chapters 12 and~21.\cr
line~16 of Chapter~6's {\tt story}\cr
lines 7 to~11\cr
lines 2,~3, 4, and~5.\cr
(2)~a big black bar\cr
All 256~characters are initially of category~12,\cr
letter~{\tt x} in family~1.\cr
the factor~$f$, where $n$~is 1000~times~$f$.\cr
\enddisplay
\answer|(cf.~Chapter~12).|\parbreak
|Chapters 12 and~21.|\parbreak
|line~16 of Chapter~6's {\tt story}|\parbreak
|lines 7 to~11|\parbreak
|lines 2,~3, 4, and~5.|\parbreak
|(2)~a big black bar|\parbreak
|All 256~characters are initially of category~12,|\parbreak
|letter~{\tt x} in family~1.|\parbreak
|the factor~$f$, where $n$~is 1000~times~$f$.|

\exercise How would you type the phrase `for all $n$ greater than
$n_0$'\thinspace?
\answer `|for all $n$~greater than~$n_0$|' avoids distracting breaks.

\exercise And how would you type `exercise 4.3.2--15'\thinspace?
\answer `|exercise \hbox{4.3.2--15}|' guarantees that there is no break
after the ^{en-dash}. But this precaution is rarely necessary, so
`|exercise 4.3.2--15|' is an acceptable answer. No |~| is needed;
`\hbox{4.3.2--15}' is so long that it causes no offense
at the beginning of a line.

\exercise Why is it better to type `|Chapter~12|' than to type
`|\hbox{Chapter 12}|'\thinspace?
\answer The space you get from |~| will stretch or shrink with the
other spaces in the same line, but the space inside an hbox has
a fixed width since that glue has already been set once and for all.
Furthermore the first alternative permits the word Chap-\break
ter to be ^{hyphenate}d.

\dangerexercise \TeX\ will sometimes break a math formula after an
equals sign. How can you stop the computer from breaking the formula
`$x=0$'\thinspace?
\answer `|\hbox{$x=0$}|' is unbreakable, and we will see later that
`|${x=0}$|' cannot be broken. Both of these solutions set the glue
surrounding the equals sign to some fixed value, but such glue normally
wants to stretch; furthermore, the |\hbox| solution might include undesirable
blank space at the beginning or end of a line, if\/ ^|\mathsurround| is
nonzero. A third solution `|$x=\nobreak0$|' avoids both defects.

\ddangerexercise Explain how you could instruct \TeX\ not to make any
breaks after explicit hyphens and dashes. \ (This is useful in
lengthy ^{bibliographies}.)
\answer |\exhyphenpenalty=10000| prohibits all such breaks, according
to the rules found later in this chapter. Similarly, |\hyphenpenalty=10000|
prevents breaks after implicit (discretionary) hyphens.

Sometimes you want to permit a line break after a `/' just as if it were
a hyphen. For this purpose plain \TeX\ allows you to say `^|\slash|';
for example, `|input\slash output|' produces `input\slash output' with
an optional break.

{\hbadness=10000
If you want to force \TeX\ to break between lines at a certain point in
^^{line breaks, forcing} ^^{breaks, forcing good}
the middle of a paragraph, just say `^|\break|'. However, that might cause the
line to be really ^^{underfull}spaced out.\break
If you want \TeX\ to fill up the right-hand part of a line
with blank space just before a forced line break,\hfil\break
without indenting the next line, say `|\hfil\break|'.\par}

\danger You may have several consecutive lines of input
for which you want the output to appear line-for-line in the same way.
One solution is to type `|\par|' at the end of each input line; but that's
somewhat of a nuisance, so plain \TeX\ provides the abbreviation
`^|\obeylines|', which causes each end-of-line in the input to be
like |\par|. After you say |\obeylines| you will get one line of
output per line of input, unless an input line ends with `|%|' or
unless it is so long that it must be broken. For example, you
probably want to use |\obeylines| if you are typesetting a ^{poem}.
Be sure to enclose |\obeylines| in a group, unless you want this
``poetry mode'' to continue to the end of your document.
\begintt
{\obeylines\smallskip
Roses are red,
\quad Violets are blue;
Rhymes can be typeset
\quad With boxes and glue.
\smallskip}
\endtt

\dangerexercise Explain the uses of\/ ^|\quad| in this poem. What would
have happened if `|\quad|' had been replaced by `^|\indent|' in both places?
\answer The second and fourth lines are indented by an additional ``quad''
of space, i.e., by one extra em in the current type style.
\ (The control sequence |\quad| does an ^|\hskip|; when \TeX\ is in
vertical mode, |\hskip| begins a new paragraph and puts glue after the
indentation.) \ If\/ |\indent| had been used instead, those lines wouldn't
have been indented any more than the first and third, because |\indent| is
implicit at the beginning of every paragraph.  Double indentation on the
second and fourth lines could have been achieved by `|\indent\indent|'.

Roughly speaking, \TeX\ breaks paragraphs into lines in the following way:
Breakpoints are inserted between words or after hyphens so as to produce lines
whose badnesses do not exceed the current ^|\tolerance|. If there's no
way to insert such breakpoints, an ^{overfull box} is set. Otherwise the
breakpoints are chosen so that the paragraph is mathematically optimal, i.e.,
best possible, in
the sense that it has no more ``^{demerits}'' than you could obtain by any
other sequence of breakpoints. Demerits are based on the badnesses of
individual lines and on the existence of such things as consecutive lines
that end with hyphens, or tight lines that occur next to loose ones.

\danger But the informal description of line breaking in the previous
paragraph is an oversimplification of what really happens. The remainder
of this chapter explains the details precisely, for people who want to
apply \TeX\ in nonstandard ways. \TeX's line-breaking algorithm
has proved to be general enough to handle a surprising variety of
different applications; this, in fact, is probably the most interesting
aspect of the whole \TeX\ system. However, every paragraph from now on
until the end of the chapter is prefaced by at least one dangerous bend
sign, so you may want to learn the following material in easy stages
instead of all at once.

\ninepoint
\danger Before the lines have been broken, a paragraph inside of \TeX\
is actually a {\sl ^{horizontal list}}, i.e., a sequence of items that
\TeX\ has gathered while in horizontal mode. We have been saying
informally that a horizontal list consists of boxes and glue; the truth
is that boxes and glue aren't the whole story. Each item in a horizontal
list is one of the following types of things:\enddanger

\smallskip
\item\bull a box (a character or ligature or rule or hbox or vbox);

\item\bull a ^{discretionary break} (to be explained momentarily);
^^{break, discretionary}

\item\bull a ``^{whatsit}'' (something special to be explained later);

\item\bull vertical material (from ^|\mark| or ^|\vadjust| or ^|\insert|);

\item\bull a glob of ^{glue} (or ^|\leaders|, as we will see later);

\item\bull a ^{kern} (something like glue that doesn't stretch or shrink);

\item\bull a ^{penalty} (representing the undesirability of breaking here);

\item\bull ``^{math-on}'' (beginning a formula) or ``^{math-off}'' (ending a
  formula).

\smallskip\noindent
The last four types (glue, kern, penalty, and math items)
are called {\sl ^{discardable}}, since they
may change or disappear at a line break; the first four types are
called non-discardable, since they always remain intact. Many of the
things that can appear in horizontal lists have not been touched on yet
in this manual, but it isn't necessary to understand them in order to
understand line breaking. Sooner or later you'll learn how each of the
gismos listed above can infiltrate a horizontal list; and if you want to
get a thorough understanding of \TeX's internal processes, you can always
use ^|\showlists| with various features of the language, in
order to see exactly what \TeX\ is doing.

\danger A discretionary break consists of three sequences of characters
called the {\sl pre-break}, {\sl post-break}, and {\sl no-break\/}
texts. The idea is that if a line break occurs here, the ^{pre-break text}
will appear at the end of the current line and the ^{post-break text} will
occur at the beginning of the next line; but if no break occurs, the
^{no-break text} will appear in the current line. Users can specify
^^|\discretionary|
discretionary breaks in complete generality by writing
\begindisplay
|\discretionary{|\<pre-break text>|}{|\<post-break text>|}{|\<no-break text>|}|
\enddisplay
where the three texts consist entirely of characters, boxes, and kerns.
For example, \TeX\ can hyphenate the word
`difficult' between the f's, even though this requires breaking the
`ffi' ligature into `f-' followed by an `fi' ligature, if the horizontal
list contains
\begintt
di\discretionary{f-}{fi}{ffi}cult.
\endtt
Fortunately you need not type such a mess yourself; \TeX's hyphenation algorithm
works behind the scenes, taking ^{ligatures} apart and putting them
into discretionary breaks when necessary.

\danger The most common case of a discretionary break is a simple
discretionary hyphen
\begintt
\discretionary{-}{}{}
\endtt
for which \TeX\ accepts the abbreviation `^|\-|'. The next most common case is
\begintt
\discretionary{}{}{}
\endtt
(an ``^{empty discretionary}''), which \TeX\ automatically inserts after
`|-|' and after every ligature that ends with `|-|'. In the case of plain
\TeX, empty discretionaries are therefore inserted after ^{hyphens} and
^{dashes}. \ (Each font has an associated ^|\hyphenchar|, which we can
assume for simplicity is equal to `|-|'.)

\danger When \TeX\ ^{hyphenates} words, it simply inserts discretionary
breaks into the horizontal list. For example, the words `|discretionary
hyphens|' are transformed into the equivalent of
\begintt
dis\-cre\-tionary hy\-phens
\endtt
if hyphenation becomes necessary. But \TeX\ doesn't apply its hyphenation
algorithm to any word that already contains a discretionary break;
therefore you can use explicit discretionaries to override \TeX's automatic
method, in an emergency.

\dangerexercise Before 1998, some ^{German} words changed their spelling
when split between lines. For example, `backen' became `bak-ken'
and `Bettuch' sometimes became `Bett-tuch'.
How can you instruct \TeX\ to produce such effects?
\answer |ba\ck/en| and |Be\ttt/uch|, where the macros |\ck/| and |\ttt/|
are defined by
\begintt
\def\ck/{\discretionary{k-}{k}{ck}}
\def\ttt/{tt\discretionary{-}{t}{}}
\endtt
The English word `eighteen' might deserve similar treatment.
\TeX's hyphenation algorithm will not make such spelling changes automatically.

\danger In order to save time, \TeX\ tries first to break a paragraph
into lines without inserting any discretionary hyphens. This first pass
will succeed if a sequence of breakpoints is found for which none
of the resulting lines has a badness exceeding the current value of
^|\pretolerance|. If the first pass fails, the method of Appendix~H is
used to hyphenate each word of the paragraph by inserting discretionary
breaks into the horizontal list, and a second attempt is
made using ^|\tolerance| instead of\/ |\pretolerance|. When the lines
are fairly wide, as they are in this manual, experiments show that
the first pass succeeds more than 90\% of the time, and that fewer than
2~words per paragraph need to be subjected to the hyphenation algorithm,
on the average. But when the lines are very narrow the
first pass usually fails rather quickly. Plain \TeX\ sets |\pretolerance=100|
and |\tolerance=200| as the default values. If you make |\pretolerance=10000|,
the first pass will essentially always succeed, so hyphenations will not
be tried (and the spacing may be terrible); on the other hand if you make
|\pretolerance=-1|, \TeX\ will omit the first pass and will try to
hyphenate immediately.

\danger Line breaks can occur only in certain places within a horizontal
list. Roughly speaking, they occur between words and after hyphens, but in
actuality they are permitted in the following five cases:\enddanger

\smallskip
\item{a)} at glue, provided that this glue is immediately preceded by a
non-discardable item, and that it is not part of a math formula (i.e., not
between math-on and math-off). A break ``at glue'' occurs at the left edge
of the glue space.

\smallskip
\item{b)} at a kern, provided that this kern is immediately followed by
glue, and that it is not part of a math formula.

\smallskip
\item{c)} at a math-off that is immediately followed by glue.

\smallskip
\item{d)} at a penalty (which might have been inserted automatically in a
formula).

\smallskip
\item{e)} at a discretionary break.

\smallskip\noindent
Notice that if two globs of glue occur next to each other, the second one
will never be selected as a breakpoint, since it is preceded by glue (which
is discardable).

\danger Each potential breakpoint has an associated ``penalty,'' which
represents the ``aesthetic cost'' of breaking at that place. In cases
(a), (b),~(c), the penalty is zero; in case~(d) an explicit penalty
has been specified; and in case~(e) the penalty is the current value of
^|\hyphenpenalty| if the pre-break text is nonempty, or the current value of
^|\exhyphenpenalty| if the pre-break text is empty.
Plain \TeX\ sets |\hyphenpenalty=50| and |\exhyphenpenalty=50|.

\danger For example, if you say `^|\penalty| |100|' at some point in a
paragraph, that position will be a legitimate place to break between
lines, but a penalty of 100 will be charged. If you say `\hbox{|\penalty-100|}'
you are telling \TeX\ that this is a rather good place to break, because
a negative penalty is really a ``^{bonus}''; a line that ends with a bonus
might even have ``merits'' (negative demerits).

\danger Any penalty that is 10000 or more is considered to be so large
^^{infinite penalty} that \TeX\ will never break there. At the other
extreme, any penalty that is $-10000$ or less is considered to be so small
that \TeX\ will always break there.  The ^|\nobreak| macro of plain \TeX\
is simply an abbreviation for `|\penalty10000|', because this prohibits a
line break. A tie in plain \TeX\ is equivalent to `|\nobreak\|\]';
there will be no break at the glue represented by |\|\] in this
case, because glue is never a legal breakpoint when it is preceded by a
discardable item like a penalty.

\dangerexercise Guess how the ^|\break| macro is defined in plain \TeX.
\answer |\def\break{\penalty-10000 }|

\dangerexercise What happens if you say |\nobreak\break| or
|\break\nobreak|?
\answer You get a forced break as if\/ |\nobreak| were not present, because
|\break| cannot be cancelled by another penalty. In general if you
have two penalties in a row, their combined effect is the same as a single
penalty whose value is the minimum of the two original values, unless
both of those values force breaks. \ (You get two breaks from
|\break\break|; the second one creates an empty line.)

\danger When a line break actually does occur, \TeX\ removes all discardable
items that follow the break, until coming to something non-discardable,
or until coming to another chosen breakpoint. For example, a sequence of
glue and penalty items will vanish as a unit, if no boxes intervene,
unless the optimum breakpoint sequence includes one or more of the penalties.
Math-on and math-off items act essentially as kerns that contribute the spacing
specified by ^|\mathsurround|; such spacing will disappear into the line
break if a formula comes at the very end or the very beginning of a line,
because of the way the rules have been formulated above.

\ddanger The ^{badness} of a line is an integer that is approximately 100
times the cube of the ratio by which the glue inside the line must stretch
or shrink to make an hbox of the required size. For example, if the line
has a total shrinkability of 10 points, and if the glue is being
compressed by a total of 9 points, the badness is computed to be~73 (since
$100\times(9/10)^3=72.9$); similarly, a line that stretches by twice its
total stretchability has a badness of 800.  But if the badness obtained by
this method turns out to be more than 10000, the value 10000 is used.  \
$\bigl($See the discussion of ``^{glue set ratio}''~$r$ and ``^{glue set
order}''~$i$ in Chapter~12; if $i\ne0$, there is infinite stretchability or
shrinkability, so the badness is zero, otherwise the badness is
approximately $\min(100r^3,10000)$.$\bigr)$ \ Overfull boxes are
considered to be infinitely bad; they are avoided whenever possible.
^^{infinite badness}

\ddanger A line whose badness is 13 or more has a glue set ratio exceeding
50\%. We call such a line {\sl ^{tight}\/} if its glue had to shrink,
{\sl ^{loose}\/} if its glue had to stretch, and {\sl ^{very loose}\/}
if it had to stretch so much that the badness is 100 or more. But if the
badness is 12 or less we say that the line is {\sl ^{decent}}. Two
adjacent lines are said to be {\sl {visually incompatible}\/} if their
classifications are not adjacent, i.e., if a tight line is next to a
loose or very loose line, or if a decent line is next to a very loose one.

\ddanger \TeX\ rates each potential sequence of breakpoints by totalling
up {\sl ^{demerits}\/} that are assessed to individual lines. The goal
is to choose breakpoints that yield the fewest total demerits. Suppose that a
line has badness~$b$, and suppose that the penalty~$p$ is associated with
the breakpoint at the end of this line. As stated above, \TeX\ will not
even consider such a line if $p\ge10000$, or if $b$~exceeds the current
tolerance or pretolerance. Otherwise the demerits of such a line are
defined by the formula
\begindisplay
$\displaystyle{d=\cases{
  (l+b)^2+p^2,&if $0\le p<10000$;\cr
  (l+b)^2-p^2,&if $-10000<p<0$;\cr
  (l+b)^2,&if $p\le-10000$.\cr}}$
\enddisplay
Here $l$ is the current value of\/ ^|\linepenalty|, a parameter that can be
increased if you want \TeX\ to try harder to keep all paragraphs to the
minimum number of lines; plain \TeX\ sets |\linepenalty=10|. For example,
a line with badness~20 ending at glue will have $(10+20)^2=900$ demerits,
if $l=10$, since there's no penalty for a break at glue. Minimizing the
total demerits of a paragraph is roughly the same as minimizing the
sum of the squares of the badnesses and penalties; this usually means
that the maximum badness of any individual line is also minimized, over
all sequences of breakpoints.

\ddangerexercise The formula for demerits has a strange discontinuity: It
seems more reasonable at first to define $d=(l+b)^2-10000^2$, in the
case $p\le-10000$. Can you account for this apparent discrepancy?
\answer Breaks are forced when $p\le-10000$, so there's no point in
subtracting a large constant whose effect on the total demerits is
known {\sl a priori}, especially when that might cause arithmetic overflow.

\ddanger Additional demerits are assessed based on pairs of adjacent lines.
If two consecutive lines are visually incompatible, in the sense explained
a minute ago, the current value of\/ ^|\adjdemerits| is added to~$d$. If two
consecutive lines end with discretionary breaks, the ^|\doublehyphendemerits|
are added. And if the second-last line of the entire paragraph ends with
a discretionary, the ^|\finalhyphendemerits| are added. Plain \TeX\ sets
up the values |\adjdemerits=10000|, |\doublehyphendemerits=10000|,
and |\finalhyphendemerits=5000|. Demerits are in units of
``badness squared,'' so the demerit-oriented parameters need to be rather
large if they are to have much effect; but tolerances and
penalties are given in the same units as badness.

\ddanger If you set ^|\tracingparagraphs||=1|, your log file will contain a
summary of \TeX's line-breaking calculations, so you can watch the tradeoffs
that occur when parameters like |\linepenalty| and |\hyphenpenalty| and
|\adjdemerits| are twiddled. The line-break data looks pretty scary at first,
but you can learn to read it with a little practice; this, in fact, is the
best way to get a solid understanding of line breaking. Here is the
trace that results from the second paragraph of the |story| file in
Chapter~6, when |\hsize=2.5in| and |\tolerance=1000|:
\begindisplay
|[]\tenrm Mr. Drofnats---or ``R. J.,'' as he pre-|\cr
|@\discretionary via @@0 b=0 p=50 d=2600|\cr
|@@1: line 1.2- t=2600 -> @@0|\cr
|ferred to be called---was hap-pi-est when |\cr
|@ via @@1 b=131 p=0 d=29881|\cr
|@@2: line 2.0 t=32481 -> @@1|\cr
|he|\cr
|@ via @@1 b=25 p=0 d=1225|\cr
|@@3: line 2.3 t=3825 -> @@1|\cr
|was at work type-set-ting beau-ti-ful doc-|\cr
|@\discretionary via @@2 b=1 p=50 d=12621|\cr
|@\discretionary via @@3 b=291 p=50 d=103101|\cr
|@@4: line 3.2- t=45102 -> @@2|\cr
|u-|\cr
|@\discretionary via @@3 b=44 p=50 d=15416|\cr
|@@5: line 3.1- t=19241 -> @@3|\cr
|ments.|\cr
|@\par via @@4 b=0 p=-10000 d=5100|\cr
|@\par via @@5 b=0 p=-10000 d=5100|\cr
|@@6: line 4.2- t=24341 -> @@5|\cr
\enddisplay
Lines that begin with `|@@|' ^^{atsign atsign} represent {\sl^{feasible
breakpoints}}, i.e., breakpoints that can be reached without any badness
exceeding the tolerance. Feasible breakpoints are numbered consecutively,
starting with |@@1|; the beginning of the paragraph is considered to be
feasible too, and it is number |@@0|. Lines that begin with `|@|' but
not `|@@|' are candidate ways to reach the feasible breakpoint that
follows; \TeX\ will select only the best candidate, when there is a choice.
Lines that do not begin with `|@|' indicate how far \TeX\ has gotten in the
paragraph. Thus, for example, we find `|@@2: line 2.0 t=32481 -> @@1|'
after `|...hap-pi-est when|' and before `|he|', so we know that feasible
breakpoint~|@@2| occurs at the space between the words |when| and |he|.
The notation `|line 2.0|' means that this feasible break comes at the end
of line~2, and that this line will be very loose. \ (The suffixes
|.0|, |.1|, |.2|, |.3| stand respectively for very loose, loose, decent,
and tight.) \ A hyphen is suffixed to the line number if that line
ends with a discretionary break, or if it is the final line of the
paragraph; for example, `|line 1.2-|' is a decent line that was hyphenated.
The notation `|t=32481|' means that the total demerits from the beginning
of the paragraph to~|@@2| are 32481, and `|-> @@1|' means that the best
way to get to |@@2| is to come from |@@1|. On the preceding line of trace
data we see the calculations for a typeset line to this point from |@@1|:
The badness is~131, the penalty is~0, hence there are 29881 demerits.
Similarly, breakpoint |@@3| presents an alternative for the second line of
the paragraph, obtained by breaking between `|he|' and `|was|'; this one
makes the second line tight, and it has only 3825 demerits when the
demerits of line~1 are added, so it appears that |@@3| will work much
better than |@@2|. However, the next feasible breakpoint (|@@4|) occurs
after `|doc-|', and the line from |@@2| to~|@@4| has only 12621 demerits,
while the line from |@@3| to~|@@4| has a whopping 103101; therefore
the best way to get from |@@0| to~|@@4| is via~|@@2|. If we regard
demerits as distances, \TeX\ is finding the ``^{shortest paths}'' from
|@@0| to each feasible breakpoint (using a variant of a well-known
algorithm for shortest paths in an acyclic graph). Finally the end of
the paragraph comes at breakpoint |@@6|, and the shortest path from
|@@0| to~|@@6| represents the best sequence of breakpoints. Following
the arrows back from~|@@6|, we deduce that~the best breaks in this
particular paragraph go through |@@5|, |@@3|, and~|@@1|.

\ddangerexercise Explain why there are 29881 demerits from |@@1| to |@@2|,
and 12621 demerits from~|@@2| to~|@@4|.
\answer $(10+131)^2+0^2+10000=29881$ and $(10+1)^2+50^2+10000=12621$.
In both cases the ^|\adjdemerits| were added because the lines were
visually incompatible (decent, then very loose, then decent); plain
\TeX's values for ^|\linepenalty| and |\adjdemerits| were used.

\ddanger If `|b=*|' ^^|*| appears in such trace data, it means that an
infeasible breakpoint had to be chosen because there was no feasible
way to keep total demerits small.

\danger We still haven't discussed the special trick that allows the
final line of a paragraph to be shorter than the others. Just before
\TeX\ begins to choose breakpoints, it does two important things: \
(1)~If the final item of the current horizontal list is glue,
^^|\unskip|
that glue is discarded. \ (The reason is that a blank space
often gets into a token list just before ^|\par| or just before |$$|,
and this blank space should not be part of the paragraph.) \ (2)~Three more
items are put at the end of the current horizontal list: |\penalty10000|
(which prohibits a line break); |\hskip\parfillskip| (which adds
``^{finishing glue}'' to the paragraph); and |\penalty-10000| (which
forces the final break). Plain \TeX\ sets ^|\parfillskip||=0pt plus1fil|,
so that the last line of each paragraph will be filled with white space
if necessary; but other settings of\/ |\parfillskip| are appropriate in
special applications. For example, the present paragraph ends flush with
the right margin, because it was typeset with |\parfillskip=0pt|;
the author didn't have to rewrite any of the text in order to make this
possible, since a long paragraph generally allows so much flexibility that
a line break can be forced at almost any point. You can have some fun
playing with paragraphs, because the algorithm for line breaking
occasionally appears to be clairvoyant. Just write paragraphs
that are long enough.\parfillskip=0pt
% the \danger macro makes this \parfillskip local!

\dangerexercise Ben ^{User} decided to say `|\hfilneg\par|' at the end of
a paragraph, intending that the negative stretchability of\/ ^|\hfilneg|
would cancel with the |\parfillskip| of plain \TeX\null. Why didn't his
bright idea work? ^^{paragraph, ending}
\answer Because \TeX\ discards a glue item that occurs just before
|\par|. Ben should have said, e.g., `|\hfilneg\ \par|'.

\dangerexercise How can you set |\parfillskip| so that the last line
of a paragraph has exactly as much white space at the right as the
first line has indentation at the left?\nobreak\hskip\parindent\hfilneg\
\answer Just say |\parfillskip|\stretch|=|\stretch|\parindent|. Of course,
\TeX\ will not be able to find appropriate line breaks unless each
paragraph is sufficiently long or sufficiently lucky; but with an
appropriate text, your output will be immaculately
symmetrical.{\parfillskip=\parindent\par}

\ddangerexercise Since \TeX\ reads an entire paragraph before it makes
any decisions about line breaks, the computer's memory capacity might
^^{capacity exceeded} be exceeded if you are typesetting the works of some
^^{Joyce, James} ^{philosopher} or modernistic novelist who writes
200-line paragraphs. Suggest a way to cope with such authors.
\answer Assuming that the author is deceased and/or set in his or her
ways, the remedy is to insert `|{\parfillskip=0pt\par\parskip=0pt\noindent}|'
in random places, after each 50 lines or so of text. \ (Every space
between words is usually a feasible breakpoint, when you get sufficiently
far from the beginning of a paragraph.)

\danger \TeX\ has two parameters called ^|\leftskip| and ^|\rightskip| that
specify glue to be inserted at the left and right of every line in a
paragraph; this glue is taken into account when badnesses and demerits are
computed.  Plain \TeX\ normally keeps |\leftskip| and |\rightskip| zero,
but it has a `^|\narrower|' macro that increases both of their values by
the current ^|\parindent|. You may want to use |\narrower| when ^{quoting}
lengthy passages from a book.
\begintt
{\narrower\smallskip\noindent
This paragraph will have narrower lines than
the surrounding paragraphs do, because it
uses the ``narrower'' feature of plain \TeX.
The former margins will be restored after
this group ends.\smallskip}
\endtt
(Try it.) \ The second `^|\smallskip|' in this example ends the paragraph.
It's important to end the paragraph before ending the group, for otherwise
the effect of\/ |\narrower| will disappear before \TeX\ begins to choose
line breaks.

\dangerexercise When an entire paragraph is typeset in ^{italic} or ^{slanted}
type, it sometimes appears to be offset on the page with respect to
other paragraphs. Explain how you could use |\leftskip| and |\rightskip|
to shift all lines of a paragraph left by $1\pt$.
\answer |{\leftskip=-1pt \rightskip=1pt| \<text> |\par}|\par
\nobreak\medskip\noindent
(This applies to a full paragraph; if you want to correct only
isolated lines, you have to do it by hand.)

\dangerexercise The ^|\centerline|, ^|\leftline|, ^|\rightline|, and ^|\line|
macros of plain \TeX\ don't take |\leftskip| and |\rightskip| into
account. How could you make them do so?
\answer `|\def\line#1{\hbox to\hsize{\hskip\leftskip#1\hskip\rightskip}}|'
is the only change needed. \ (Incidentally,
^{displayed equations} don't take account of\/ |\leftskip| and |\rightskip|
either; it's more difficult to change that, because so many variations
are possible.)

\ddanger If you suspect that ^|\raggedright| setting is accomplished by
some appropriate manipulation of\/ |\rightskip|, you are correct. But some
care is necessary. For example, a person can
set |\rightskip=0pt plus1fil|, and every
line will be filled with space at the right. But this isn't a particularly
good way to make ragged-right margins, because the infinite stretchability
will assign zero badness to lines that are very short.
To do a decent job of ragged-right setting, the trick is to set
|\rightskip| so that it will stretch enough to make line breaks possible,
yet not too much, because short lines should be considered bad. Furthermore
the spaces between words should be fixed so that they do not stretch
or shrink. \ (See the definition of\/ |\raggedright| in Appendix~B\null.) \
It would also be possible to allow a little variability in the interword
glue, so that the right margin would not be quite so ragged but the
paragraphs would still have an informal appearance.

\danger \TeX\ looks at the parameters that affect
line breaking only when it is breaking lines. For example, you shouldn't
try to change the ^|\hyphenpenalty| in the middle of a paragraph,
if you want \TeX\ to penalize the hyphens in one word more than it does in
another word. The relevant values of
|\hyphenpenalty|, |\rightskip|, |\hsize|, and so on,
are the ones that are current at the end of the paragraph.
On the other hand, the width of indentation that you get
implicitly at the beginning of a paragraph or when you say `^|\indent|'
is determined by the value of\/ ^|\parindent| at the time the
indentation is contributed to the current horizontal list,
not by its value at the end of the paragraph. Similarly, penalties
that are inserted into math formulas within a paragraph are based
on the values of\/ ^|\binoppenalty| and ^|\relpenalty| that are current
at the end of each particular formula. Appendix~D contains an example
that shows how to have both ragged-right and ragged-left margins within
a single paragraph, without using |\leftskip| or |\rightskip|.

\varunit=0.989pt % getting ready to make circular insert
% \varunit=1.078pt was used with amr5: it had more letterspacing
\setbox0=\vtop{\null
\baselineskip6\varunit
\parfillskip0pt
\parshape 19
-18.25\varunit 36.50\varunit
-30.74\varunit 61.48\varunit
-38.54\varunit 77.07\varunit
-44.19\varunit 88.39\varunit
-48.47\varunit 96.93\varunit
-51.70\varunit 103.40\varunit
-54.08\varunit 108.17\varunit
-55.72\varunit 111.45\varunit
-56.68\varunit 113.37\varunit
-57.00\varunit 114.00\varunit
-56.68\varunit 113.37\varunit
-55.72\varunit 111.45\varunit
-54.08\varunit 108.17\varunit
-51.70\varunit 103.40\varunit
-48.47\varunit 96.93\varunit
-44.19\varunit 88.39\varunit
-38.54\varunit 77.07\varunit
-30.74\varunit 61.48\varunit
-18.25\varunit 36.50\varunit
\fiverm
\frenchspacing
\noindent
\hbadness 6000
\tolerance 9999
\pretolerance 0
\hyphenation{iso-peri-met-ric}
The area of a circle is a mean proportional
between any two regular and similar polygons of which one
circumscribes it and the other is isoperimetric with it.
In addition, the area of the circle is less than that of any
circumscribed polygon and greater than that of any
isoperimetric polygon. And further, of these
circumscribed polygons, the one that has the greater number of sides
has a smaller area than the one that has a lesser number;
but, on the other hand, the isoperimetric polygon that
has the greater number of sides is the larger.
\hbox to 36.50\varunit{\hss[Galileo,\thinspace1638]\hss}
}

\danger
\parshape 16
3pc 26pc
3pc 26pc
0pc 24.69pc
0pc 23.51pc
0pc 22.73pc
0pc 22.20pc
0pc 21.85pc
0pc 21.65pc
0pc 21.58pc
0pc 21.65pc
0pc 21.85pc
0pc 22.20pc
0pc 22.73pc
0pc 23.51pc
0pc 24.69pc
0pc 29pc
\vadjust{\moveright 28pc\vbox to 0pt{\vskip88pt\vskip-60\varunit
  \vskip-3pt\box0\vss}}%
\strut It's possible to control the length of lines in a much more general
way, if simple changes to |\leftskip| and |\rightskip| aren't
flexible enough for your purposes. For example, a semicircular
^{hole} has been cut out of the present paragraph, in order to make
room for a circular illustration that contains some of ^{Galileo}'s
immortal words about ^{circle}s; all of the line breaks in this paragraph
and in the circular quotation were found by \TeX's line-breaking
algorithm. You can specify an essentially arbitrary paragraph
shape by saying ^|\parshape||=|\<number>, where the \<number> is
a positive integer $n$, followed by $2n$ \<dimen> specifications.
In general, `|\parshape=|$n$ $i_1$~$l_1$ $i_2$~$l_2$ $\ldots$ $i_n$~$l_n$'
specifies a paragraph whose first $n$ lines will have lengths
$l_1$, $l_2$, \dots,~$l_n$, respectively, and they will be
indented from the left margin by the respective amounts
$i_1$, $i_2$, \dots,~$i_n$. If the paragraph has fewer than
$n$ lines, the additional specifications will be ignored;
if it has more than $n$ lines, the specifications for line $n$ will
be repeated ad infinitum. You can cancel the effect of a previously
specified |\parshape| by saying `|\parshape=0|'.\parfillskip0pt
 ^^{illustrations, fitting copy around}

\ddangerexercise Typeset the following ^{Pascal}ian quotation in
the shape of an isosceles ^{triangle}:
``I turn, in the following treatises, to various uses of those
    triangles whose generator is unity. But I leave out many more
    than I include; it is extraordinary how fertile in properties
    this triangle is. Everyone can try his hand.''
\answer The author's best solution is based on a variable |\dimen|
register |\x|:
\begintt
\setbox1=\hbox{I}
\setbox0=\vbox{\parshape=11 -0\x0\x -1\x2\x -2\x4\x -3\x6\x
   -4\x8\x -5\x10\x -6\x12\x -7\x14\x -8\x16\x -9\x18\x -10\x20\x
  \ifdim \x>2em \rightskip=-\wd1
  \else \frenchspacing \rightskip=-\wd1 plus1pt minus1pt
   \leftskip=0pt plus 1pt minus1pt \fi
  \parfillskip=0pt \tolerance=1000 \noindent I turn, ... hand.}
\centerline{\hbox to \wd1{\box0\hss}}
\endtt
Satisfactory results are obtained with font |cmr10| when |\x| is set to
$8.9\pt$, $13.4\pt$, $18.1\pt$, $22.6\pt$, $32.6\pt$, and $47.2\pt$,
yielding triangles that are respectively 11,~9, 8, 7, 6, and~5 lines tall.

\danger You probably won't need unusual parshapes very often. But there's
a special case that occurs rather frequently, so \TeX\ provides a special
abbreviation for~it in terms of two parameters called ^|\hangindent| and
^|\hangafter|. The command `|\hangindent=|\<dimen>' specifies a so-called
^{hanging indentation}, and the command `|\hangafter=|\<number>' specifies
the duration of that indentation. Let $x$ and $n$ be the respective values
of\/ |\hangindent| and |\hangafter|, and let $h$ be the value of\/
^|\hsize|; then if $n\ge0$, hanging indentation will occur on lines $n+1$,
$n+2$, $\ldots$ of the paragraph, but if $n<0$ it will occur on lines
1,~2, \dots,~$\vert n\vert$. Hanging indentation means that lines will be
of width $h-\vert x\vert$ instead of their normal width~$h$; if $x\ge0$,
the lines will be indented at the left margin, otherwise they will be
indented at the right margin. For example, the ``dangerous bend''
paragraphs of this manual have a hanging indentation of 3~picas that lasts
for two lines; they were set with |\hangindent=3pc| and |\hangafter=-2|.

\danger Plain \TeX\ uses hanging indentation in its `^|\item|' macro, which
produces a paragraph in which every line has the same indentation as a
normal |\indent|. Furthermore, |\item| takes a parameter that is placed
into the position of the indentation on the first line. Another macro called
`^|\itemitem|' does the same thing but with double indentation.
For example, suppose you type
\begintt
\item{1.} This is the first of several cases that are being
enumerated, with hanging indentation applied to entire paragraphs.
\itemitem{a)} This is the first subcase.
\itemitem{b)} And this is the second subcase. Notice
that subcases have twice as much hanging indentation.
\item{2.} The second case is similar.
\endtt
{\let\par=\endgraf Then you get the following output:
\medskip
\item{1.} This is the first of several cases that are being
enumerated, with hanging indentation applied to entire paragraphs.
\itemitem{a)} This is the first subcase.
\itemitem{b)} And this is the second subcase. Notice
that subcases have twice as much hanging indentation.
\item{2.} The second case is similar.
\medskip}\noindent\hangindent0pt
(Indentations in plain \TeX\ are not actually as dramatic as those
displayed here; Appendix~B
says `|\parindent=20pt|', but this manual has been set with
|\parindent=36pt|.) \ It is customary to put ^|\medskip| before and after
a group of itemized paragraphs, and to say |\noindent|
before any closing remarks that apply to all of the cases.
^^{enumerated cases in separate paragraphs}
Blank lines are not needed before |\item| or |\itemitem|, since those macros
begin with |\par|.

\dangerexercise Suppose one of the enumerated cases continues for two
or more paragraphs. How can you use |\item| to get hanging indentation
on the subsequent paragraphs?
\answer |\item{}| at the beginning of each paragraph that wants hanging
indentation.

\dangerexercise Explain how to make a ``^{bullet}ed'' item that says `$\bullet$'
instead of `1.'.
\answer |\item{$\bullet$}|

\ddangerexercise The `|\item|' macro doesn't alter the right-hand margin. How
could you indent at both sides?
\answer Either change |\hsize| or |\rightskip|. The trick is to change it back
again at the end of a paragraph. Here's one way, without grouping:
\begintt
\let\endgraf=\par \edef\restorehsize{\hsize=\the\hsize}
\def\par{\endgraf \restorehsize \let\par=\endgraf}
\advance\hsize by-\parindent
\endtt

\ddangerexercise Explain how you could specify a hanging indentation
of $-2$ ems (i.e., the lines should project into the left margin),
after the first two lines of a paragraph.
\answer |\dimen0=\hsize \advance\dimen0 by 2em|\parbreak
|\parshape=3 0pt\hsize 0pt\hsize -2em\dimen0|

\danger If\/ |\parshape| and hanging indentation have both been specified,
|\parshape| takes precedence and |\hangindent| is ignored. You get the
normal paragraph shape, in which every line width is |\hsize|, when
|\parshape=0|, |\hangindent=0pt|, and |\hangafter=1|. \TeX\ automatically
restores these normal values at the end of every paragraph, and (by
local definitions) whenever it enters internal vertical mode. For example,
hanging indentation that might be present outside of a ^|\vbox| construction
won't occur inside that vbox, unless you ask for it inside.
^^{paragraph shape reset} ^^{hanging indentation reset}

\ddangerexercise Suppose you want to leave room at the right margin for
a rectangular illustration that takes up 15 lines, and you expect that
three paragraphs will go by before you have typeset enough text to get
past that illustration. Suggest a good way to do this without trial and error,
given the fact that \TeX\ resets hanging indentation.
\answer The three paragraphs can be combined into a single paragraph, if
you use `|\hfil\vadjust{\vskip\parskip}\break\indent|' instead of
`|\par|' after the first two.  Then of course you say, e.g.,
|\hangindent=-50pt \hangafter=-15|. \ (The same idea can be applied in
connection with |\looseness|, if you want \TeX\ to make one of three
paragraphs looser but if you don't want to choose which one it will be.
However, long paragraphs fill \TeX's memory; please use restraint.) \
See also the next exercise.

\ddanger If ^{displayed equations} occur in a paragraph that has a nonstandard
shape, \TeX\ always assumes that the display takes up exactly three lines.
For example, a paragraph that has four lines of text, then a display, then
two more lines of text, is considered to be $4+3+2=9$ lines long; the
displayed equation will be indented and centered using the paragraph shape
information appropriate to line~6.

\ddanger \TeX\ has an internal integer variable called ^|\prevgraf| that
records the number of lines in the most recent paragraph that has been
completed or partially completed. You can use |\prevgraf| in the context of
a \<number>, and you can set |\prevgraf| to any desired nonnegative value
if you want to make \TeX\ think that it is in some particular part of the
current paragraph shape. For example, let's consider again a paragraph
that contains four lines plus a display plus two more lines. When \TeX\
starts the paragraph, it sets |\prevgraf=0|; when it starts the display,
|\prevgraf| will be~4; when it finishes the display, |\prevgraf| will
be~7; and when it ends the paragraph, |\prevgraf| will be~9. If the
display is actually one line taller than usual, you could set
|\prevgraf=8| at the beginning of the two final lines; then \TeX\ will
think that a 10-line paragraph is being made. The value of\/ |\prevgraf|
affects line breaking only when \TeX\ is dealing with nonstandard
|\parshape| or |\hangindent|.

\edef\lastex{\chapno.\the\exno}
\ddangerexercise Solve exercise \lastex\ using |\prevgraf|.
\answer Use |\hangcarryover| between paragraphs, defined as follows:
\begintt
\def\hangcarryover{\edef\next{\hangafter=\the\hangafter
    \hangindent=\the\hangindent}
  \par\next
  \edef\next{\prevgraf=\the\prevgraf}
  \indent\next}
\endtt

\ddanger You are probably convinced by now that \TeX's line-breaking algorithm
has plenty of bells and whistles, perhaps even too many.
But there's one more feature,
called ``looseness''; some day you might find yourself needing it,
when you are fine-tuning the pages of a book. If you set |\looseness=1|,
\TeX\ will try to make the current paragraph one line longer than its
optimum length, provided that there is a way to choose such breakpoints
without exceeding the tolerance you have specified for the badnesses
of individual lines. Similarly, if you set |\looseness=2|, \TeX\ will
try to make the paragraph two lines longer; and |\looseness=-1| causes an
attempt to make it shorter. The general idea is that \TeX\ first finds
breakpoints as usual; then if the optimum breakpoints produce
$n$~lines, and if the current ^|\looseness| is~$l$, \TeX\ will choose
the final breakpoints so as to make the final number of lines as close
as possible to $n+l$ without exceeding the current tolerance. Furthermore,
the final breakpoints will have fewest total demerits, considering all ways
to achieve the same number of~lines.

\ddanger For example, you can set |\looseness=1| if you want to avoid
a lonely ``^{club line}'' or ``^{widow line}'' on some page that does not
have sufficiently flexible glue, or if you want the total number of
lines in some two-column document to come out to be an even number.
It's usually best to choose a paragraph that is already pretty ``full,''
i.e., one whose last line doesn't have much white space, since such
paragraphs can generally be loosened without much harm. You might
also want to insert a ^{tie} between the last two words of that paragraph,
so that the loosened version will not end with only one ``^{widow word}'' on the
^^{orphans, see widow words}
line; this tie will cover your tracks, so that people will find it hard to
detect the fact that you have tampered with the spacing. On the other
hand, \TeX\ can take almost any sufficiently long paragraph and stretch
it a bit, without substantial harm; the present paragraph is, in fact,
one line looser than \hbox{its optimum length}.\looseness=1

\ddanger \TeX\ resets the looseness to zero at the same time as it resets
|\hangindent|, |\hangafter|, and |\parshape|.

\ddangerexercise Explain what \TeX\ will do if you set |\looseness=-1000|.
\answer It will set the current paragraph in the minimum number of lines
that can be achieved without violating the tolerance; and, given that
number of lines, it will break them optimally. \ (However, nonzero
looseness makes \TeX\ work harder, so this is not recommended if you
don't want to pay for the extra computation. You can achieve almost the
same result much more efficiently by setting ^|\linepenalty||=100|, say.)

\danger Just before switching to horizontal mode to begin scanning a
paragraph, \TeX\ inserts the glue specified by ^|\parskip| into the vertical
list that will contain the paragraph, unless that vertical list is empty so
far. For example, `|\parskip=3pt|' will cause 3~points of extra space
to be placed between paragraphs. Plain \TeX\ sets |\parskip=0pt plus1pt|;
this gives a little stretchability, but no extra space.

\danger After line breaking is complete, \TeX\ appends the lines to the
current vertical list that encloses the current paragraph, inserting
interline glue as explained in Chapter~12; this interline glue will
depend on the values of\/ ^|\baselineskip|, ^|\lineskip|, and ^|\lineskiplimit|
that are currently in force. \TeX\ will also insert penalties into
the vertical list, just before each glob of ^{interline glue}, in order to
help control page breaks that might have to be made later. For example, a
special penalty will be assessed for breaking a page between the first two
lines of a paragraph, or just before the last line, so that ``club'' or
``widow'' lines that are detached from the rest of a paragraph will not
appear all alone on a page unless the alternative is worse.

\danger Here's how interline penalties are calculated: \TeX\ has just
chosen the breakpoints for some paragraph, or for some partial paragraph
that precedes a displayed equation; and $n$~lines have been formed.
The penalty between lines $j$ and $j+1$, given a value of $j$ in the
range $1\le j<n$, is the value of\/ ^|\interlinepenalty| plus
additional charges made in special cases: The ^|\clubpenalty| is
added if $j=1$, i.e., just after the first line; then the
^|\displaywidowpenalty| or the ^|\widowpenalty| is added if $j=n-1$,
i.e., just before the last line, depending on whether or not
the current lines immediately precede a display; and finally the
^|\brokenpenalty| is added, if the $j$th line ended at a discretionary break.
(Plain \TeX\ sets |\clubpenalty=150|, |\widowpenalty=150|,
|\displaywidowpenalty=50|, and |\brokenpenalty=100|; the value of\/
|\interlinepenalty| is normally zero, but it is increased to 100 within
^{footnotes}, so that long footnotes will tend not to be broken between
pages.)

\dangerexercise Consider a five-line paragraph in which the second and fourth
lines end with hyphens. What penalties does  plain \TeX\ put between the lines?
\answer 150, 100, 0, 250. \ (When the total penalty is zero, as between lines
3 and~4 in this case, no penalty is actually inserted.)

\dangerexercise What penalty goes between the lines of a two-line paragraph?
\answer |\interlinepenalty| plus |\clubpenalty| plus |\widowpenalty| (and
also plus |\brokenpenalty|, if the first line ends with a discretionary break).

\ddanger If you say ^|\vadjust||{|\<vertical mode material>|}| within a
paragraph, \TeX\ will use internal vertical mode to insert the specified
material into the vertical
list that encloses the paragraph, immediately after whatever line
contained the position of the |\vadjust|. For example, you can say
`|\vadjust{\kern1pt}|' to increase the amount of space between lines of a
paragraph if those lines would otherwise come out too close together.  \ (The
\vadjust{\kern1pt}author
did that in the current line, just to illustrate what happens.) \ Also,
if you want to make sure that a page break will occur immediately after a
certain line, you can say `|\vadjust{\eject}|' ^^|\eject| anywhere in that line.

\ddanger Later chapters discuss |\insert| and |\mark| commands that are
relevant to \TeX's page builder. If such commands appear within a
paragraph, they are removed from whatever horizontal lines contain them
and placed into the enclosing vertical list, together with other vertical
material from |\vadjust| commands that might be present. In the final
vertical list, each horizontal line of text is an hbox that is immediately
preceded by interline glue and immediately followed by vertical material
that has ``^{migrate}d out'' from that line (with left to right order
preserved, if there are several instances of vertical material); then
comes the interline penalty, if it is nonzero. Inserted vertical material
does not influence the ^{interline glue}.

\ddangerexercise Design a |\marginalstar| macro ^^{marginal notes}
that can be used anywhere in a paragraph. It should use |\vadjust| to
place an asterisk in the margin just to the left of the line where
|\marginalstar| occurs.
\answer The tricky part is to avoid ``opening up'' the paragraph by
adding anything to its height; yet this star is to be contributed after
a line having an unknown depth, because the depth of the line depends
on details of line breaking that aren't known until afterwards.
The following solution uses ^|\strut|, and assumes that the line containing
the marginal star does not have depth exceeding ^|\dp||\strutbox|, the
depth of a ^|\strut|.
\begintt
\def\strutdepth{\dp\strutbox}
\def\marginalstar{\strut\vadjust{\kern-\strutdepth\specialstar}}
\endtt
(The next line must also not be too tall.)
Here |\specialstar| is a box of height zero and depth |\strutdepth|,
and it puts an asterisk in the left margin:
\begintt
\def\specialstar{\vtop to \strutdepth{
    \baselineskip\strutdepth
    \vss\llap{* }\null}}
\endtt

\ddanger When \TeX\ enters ^{horizontal mode}, it will interrupt its normal
scanning to read tokens that were predefined by the command
^|\everypar||={|\<token list>|}|. For example, suppose you have said
`|\everypar={A}|'. If you type `|B|' in vertical mode, \TeX\ will shift
to horizontal mode (after contributing ^|\parskip| glue to the current
page), and a horizontal list will be initiated by inserting an empty box
of width ^|\parindent|. Then \TeX\ will read `|AB|', since it reads the
|\everypar| tokens before getting back to the `|B|' that triggered the
new paragraph. Of course, this is not a very useful illustration of
|\everypar|; but if you let your imagination run you will think of
better applications.

\ddangerexercise Use |\everypar| to define an |\insertbullets| macro: All
paragraphs in a group of the form `|{\insertbullets ...\par}|' should have a
bullet symbol `$\bullet$' as part of their indentation.
^^{bulleted lists}
\answer |\def\insertbullets{\everypar={\llap{$\bullet$\enspace}}}|\par
\nobreak\smallskip\noindent
(A similar device can be used to insert hanging indentation,
and/or to number the paragraphs automatically.)

\ddanger A paragraph of zero lines is formed if you say `|\noindent\par|'.
If\/ |\everypar| is null, such a paragraph contributes nothing except
|\parskip| glue to the current vertical list.

\ddangerexercise Guess what happens if you say `|\noindent$$...$$ \par|'.
\answer First comes |\parskip| glue (but you might not see it on the current
page if you say |\showlists|, since glue disappears at the top of each
page). Then comes the result of\/ |\everypar|, but let's assume that
|\everypar| doesn't add anything to the horizontal list, so that
you get an empty horizontal list; then there's no partial paragraph
before the display. The displayed equation follows the normal rules
(it occupies lines 1--3 of the paragraph, and uses the indentation and
length of line~2, if there's a nonstandard shape).  Nothing follows the
display, since a blank space is ignored after a closing `|$$|'.\par
Incidentally, the behavior is different if you start a paragraph with
`|$$|' instead of with |\noindent$$|, ^^{display at beginning of paragraph}
since \TeX\ inserts a paragraph indentation that will appear on a line by
itself (with |\leftskip| and |\parfillskip| and |\rightskip| glue).

\ddanger Experience has shown that \TeX's line-breaking algorithm can be
harnessed to a surprising variety of tasks. Here, for example, is an application
that indicates one of the possibilities: Articles that are published in
{\sl^{Mathematical Reviews}\/\null} are generally signed with the reviewer's
name and address, and this information is typeset flush right, i.e., at
the right-hand margin. ^^{flush right}
If there is sufficient space to put such a name and address at the right of
the final line of the paragraph, the publishers can save space, and at the same
time the results look better because there are no strange gaps on the page.
\def\signed #1 (#2){{\unskip\nobreak\hfil\penalty50\hskip2em
  \hbox{}\nobreak\hfil\sl#1\/ \rm(#2)
  \parfillskip=0pt \finalhyphendemerits=0 \endgraf}}
\begindisplay
\vbox{\hsize 3.0in \parindent0pt
  This is a case where the name and address fit in nicely with the review.
  \signed A. Reviewer (Ann Arbor, Mich.)
  \medskip
  But sometimes an extra line must be added. \signed N. Bourbaki (Paris)}
\enddisplay
^^{Reviewer} ^^{Bourbaki}
Let's suppose that a space of at least two ems should separate the reviewer's
name from the text of the review, if they occur on the same line. We would
like to design a macro so that the examples shown above could be typed
as follows in an input file:
\begintt
... with the review. \signed A. Reviewer (Ann Arbor, Mich.)
... an extra line must be added. \signed N. Bourbaki (Paris)
\endtt
Here is one way to solve the problem:
\begintt
\def\signed #1 (#2){{\unskip\nobreak\hfil\penalty50
  \hskip2em\hbox{}\nobreak\hfil\sl#1\/ \rm(#2)
  \parfillskip=0pt \finalhyphendemerits=0 \par}}
\endtt
If a line break occurs at the |\penalty50|, the |\hskip2em| will disappear
and the empty |\hbox| will occur at the beginning of a line, followed by
|\hfil| glue. This yields two lines whose badness is zero; the first of these
lines is assessed a penalty of~50. But if no line break occurs at the
|\penalty50|, there will be glue of $2\em$ plus $2\,{\rm fil}$ between
the review and the name; this yields one line of badness zero. \TeX\ will
try both alternatives, to see which leads to the fewest total demerits.
The one-line solution will usually be preferred if it is feasible.

\ddangerexercise Explain what would happen if `|\hbox{}|' were left out
of the ^|\signed| macro.
\answer A break at |\penalty50| would cancel |\hskip2em\nobreak\hfil|,
so the next line would be forced to start with the reviewer's name flush left.
\ (But ^|\vadjust||{}| would actually be better than |\hbox{}|; it
uses \TeX\ more efficiently.)

\ddangerexercise Why does the |\signed| macro say
`^|\finalhyphendemerits||=0|'\thinspace?
\answer Otherwise the line-breaking algorithm might prefer two final lines to
one final line, simply in order to move a hyphen from the second-last line up
to the third-last line where it doesn't cause demerits. This in fact caused
some surprises when the |\signed| macro was being tested; |\tracingparagraphs=1|
was used to diagnose the problem.

{\hbadness=10000
\ddangerexercise In one of the paragraphs earlier in this chapter, the author
used ^|\break| to force a line break in a specific place; as a result, the
third line of that particular paragraph was really spaced out.\break
Explain why all the extra space went into the third line, instead of being
distributed impartially among the first three lines.
\answer Distributing the extra space evenly would lead to three lines of
the maximum badness (10000). It's better to have just one bad line
instead of three, since \TeX\ doesn't distinguish degrees of badness when
lines are really awful. In this particular case the ^|\tolerance| was 200,
so \TeX\ didn't try any line breaks that would stretch the first two lines;
but even if the tolerance had been raised to 10000, the optimum setting would
have had only one underfull line. If you really want to spread the
space evenly you can do so by using ^|\spaceskip| to increase the
amount of stretchability between words.

}
\ddanger If you want to avoid overfull boxes at all costs without
trying to fix them manually, you might be tempted to set
|\tolerance=10000|; this allows arbitrarily bad lines to be acceptable
in tough situations. But infinite tolerance is a bad idea, because
\TeX\ doesn't distinguish between terribly bad and preposterously
horrible lines. Indeed, a tolerance of 10000 encourages \TeX\ to
concentrate all the badness in one place, making one truly unsightly
line instead of two moderately bad ones, because a single
``write-off'' produces fewest total demerits according to the rules.
There's a much better way to get the desired effect: \TeX\ has a
parameter called ^|\emergencystretch| that is added to the assumed
stretchability of every line when badness and demerits are computed,
in cases where overfull boxes are otherwise unavoidable. If
|\emergencystretch| is positive, \TeX\ will make a third pass over a
paragraph before choosing the line breaks, when the first passes did
not find a way to satisfy the ^|\pretolerance| and ^|\tolerance|.
The effect of\/ |\emergencystretch| is to scale down the badnesses so
that large infinities are distinguishable from smaller ones. By
setting |\emergencystretch| high enough (based on |\hsize|) you can be
sure that the |\tolerance| is never exceeded; hence overfull boxes
will never occur unless the line-breaking task is truly impossible.

\ddangerexercise Devise a ^|\raggedcenter| macro (analogous to ^|\raggedright|)
that partitions the words of a paragraph into as few as possible lines
of approximately equal size and centers each individual line. Hyphenation
should be avoided if possible.
\answer |\def\raggedcenter{\leftskip=0pt plus4em \rightskip=\leftskip|%
\parbreak|\parfillskip=0pt \spaceskip=.3333em \xspaceskip=.5em|\parbreak
        |\pretolerance=9999 \tolerance=9999 \parindent=0pt|\parbreak
        |\hyphenpenalty=9999 \exhyphenpenalty=9999 }|

\endchapter

When the author objects to [a hyphenation]\/
he should be asked to add or cancel or substitute
a word or words that will prevent the breakage.
\smallskip
Authors who insist on even spacing always,
with sightly divisions always,
do not clearly understand the rigidity of types.
\author T. L. ^{DE VINNE}, {\sl Correct Composition\/} (1901) % p138, p206

\bigskip

In reprinting his own works, whenever [William ^{Morris}]\/
found a line that justified awkwardly, he altered the wording
solely for the sake of making it look well in print.
\smallskip
When a proof has been sent me with two or three
lines so widely spaced as to make a grey band across the page,
I have often rewritten the passage so as to fill up the lines better;
but I am sorry to say that my object has generally been so little
understood that the compositor has spoilt all the rest
of the paragraph instead of mending his former bad work.
\author GEORGE BERNARD ^{SHAW}, in {\sl The Dolphin\/} (1940) % v4.1 p80

\eject
\beginchapter Chapter 15. How \TeX\ Makes\\Lines into Pages

\tracingpages=1
\TeX\ attempts to choose desirable places to divide your document into
individual pages, and its technique for doing this usually works pretty
well. But the problem of ^{page make-up} is considerably more difficult
than the problem of line breaking that we considered in the previous chapter,
because pages often have much less flexibility than lines do. If the
vertical glue on a page has little or no ability to stretch or to shrink,
\TeX\ usually has no choice about where to start a new page; conversely, if
there is too much variability in the glue, the result will look bad because
different pages will be too irregular. Therefore if you are fussy about the
appearance of pages, you can expect to do some rewriting of the manuscript
until you achieve an appropriate balance, or you might need to fiddle
with the ^|\looseness| as described in Chapter~14; no automated system will
be able to do this as well as you.

Mathematical papers that contain a lot of displayed equations have an
advantage in this regard, because the glue that surrounds a display tends to
be quite flexible. \TeX\ also gets valuable room to maneuver when you
have occasion to use ^|\smallskip| or ^|\medskip| or ^|\bigskip| spacing
between certain paragraphs. For example, consider a page that contains
a dozen or so exercises, and suppose that there is $3\pt$ of additional
space between exercises, where this space can stretch to $4\pt$ or
shrink to $2\pt$. Then there is a chance to squeeze an extra line on the page,
or to open up the page by removing one line, in order to avoid splitting
an exercise between pages. Similarly, it is possible to use flexible
glue in special publications like membership rosters or company telephone
directories, so that individual entries need not be split between columns
or pages, yet every column appears to be the same height.

For ordinary purposes you will probably find that \TeX's automatic method
of page breaking is satisfactory. And when it occasionally gives
unpleasant results, you can force the machine to break at your favorite
place by typing `^|\eject|'. But be careful: |\eject| will cause \TeX\ to
stretch the page out, if necessary, so that the top and bottom baselines
agree with those on other pages. If you want to eject a short page,
filling it with blank space at the bottom, type `|\vfill\eject|' instead.

\danger If you say `|\eject|' in the middle of a paragraph, the paragraph
will end first, as if you typed `|\par\eject|'. But Chapter~14 mentions
that you can say `^|\vadjust||{\eject}|' in mid-paragraph, if you want to
force a page break after whatever line contains your current position
when the full paragraph is eventually broken up into lines; the rest of the
paragraph will go on the following page.

\danger To prevent a page break, you can say `^|\nobreak|' in vertical
mode, just as |\nobreak| in horizontal mode prevents breaks between lines.
For example, it is wise to say |\nobreak| between the title of a subsection
and the first line of text in that subsection. But |\nobreak| does not
cancel the effect of other commands like |\eject| that tell \TeX\ to
break; it only inhibits a break at glue that immediately follows. You
should become familiar with \TeX's rules for line breaks and page breaks
if you want to maintain fine control over everything. The remainder of
this chapter is devoted to the intimate details of page breaking.

\ninepoint
\danger \TeX\ breaks lists of lines into pages by computing badness ratings
and penalties, more or less as it does when breaking paragraphs into lines.
But pages are made up one at a time and removed from \TeX's memory; there is
no looking ahead to see how one page break will affect the next one.
In other words, \TeX\ uses a special method to find the optimum
breakpoints for the lines in an entire paragraph, but it doesn't attempt
to find the optimum breakpoints for the pages in an entire document. The
computer doesn't have enough high-speed memory capacity to remember the
contents of several pages, so \TeX\ simply chooses each page break as best
it can, by a process of ``local'' rather than ``global'' optimization.

\tracingpages=0
\danger Let's look now at the details of \TeX's page-making process.
Everything you contribute to the pages of your document is placed on
the {\sl ^{main vertical list}}, which is the sequence of items that
\TeX\ has accumulated while in vertical mode. Each item in a ^{vertical
list} is one of the following types of things:\enddanger

\smallskip
\item\bull a box (an hbox or vbox or rule);

\item\bull a ``^{whatsit}'' (something special to be explained later);

\item\bull a ^{mark} (another thing that will be explained later);

\item\bull an ^{insertion} (yet another thing that we will get to);

\item\bull a glob of ^{glue} (or ^|\leaders|, as we will see later);

\item\bull a ^{kern} (something like glue that doesn't stretch or shrink);

\item\bull a ^{penalty} (representing the undesirability of breaking here).

\smallskip\noindent
The last three types (glue, kern, and penalty items) are called ^{discardable},
for the same reason that we called them discardable in horizontal lists.
You might want to compare these specifications with the analogous rules
for the horizontal case, found in Chapter~14; it turns out that vertical
lists are just like horizontal ones except that character boxes, discretionary
breaks, ^|\vadjust| items, and math shifts cannot appear in vertical lists.
Chapter~12 exhibits a typical vertical list in \TeX's internal box-and-glue
representation.

\danger Page breaks can occur only at certain places within a vertical list.
The permissible breakpoints are exactly the same as in the horizontal case,
namely\enddanger

\smallskip
\item{a)} at glue, provided that this glue is immediately preceded by a
non-discardable item (i.e., by a box, whatsit, mark, or insertion);

\smallskip
\item{b)} at a kern, provided that this kern is immediately followed by glue;

\smallskip
\item{c)} at a penalty (which might have been inserted automatically in a
paragraph).

\smallskip\noindent
Interline glue is usually inserted automatically between the boxes of a
vertical list, as explained in Chapter~12, so there is usually a valid
breakpoint between boxes.

\danger As in horizontal lists, each potential breakpoint has an
associated penalty, which is high for undesirable breakpoints and negative
for desirable ones. The penalty is zero at glue and kern breaks, so it is
nonzero only at explicit penalty breaks.  If you say `^|\penalty||-100|'
between two paragraphs, you are indicating that \TeX\ should try to break
here because the penalty is negative; a bonus of 100 points for breaking
at this place will essentially cancel up to 100 units of badness that
might be necessary to achieve such a break. A penalty of 10000 or more is
so large that it inhibits breaking; a penalty of $-10000$ or less is so
small that it forces breaking.

\danger Plain \TeX\ provides several control sequences that help to
control page breaks. For example, ^|\smallbreak|, ^|\medbreak|, and
^|\bigbreak| specify increasingly desirable places to break, having
respective penalties of $-50$, $-100$, and~$-200$; furthermore, they will
insert a ^|\smallskip|, ^|\medskip|, or ^|\bigskip| of space,
respectively, if a break is not taken. However, |\smallbreak|,
|\medbreak|, and |\bigbreak| do not increase existing glue unnecessarily;
for example, if you say |\smallbreak| just after a displayed
equation, you won't get a |\smallskip| of space in addition to the glue that
already follows a display. Therefore these commands can conveniently
be used before and after the statements of ^{theorems}, in a format for
mathematical papers. In the present manual the author has used a macro
that puts |\medbreak| before and after every dangerous-bend paragraph;
|\medbreak\medbreak| is equivalent to a single |\medbreak|, so you don't
see two medskips when one such paragraph ends and another one~begins.

\danger The ^|\goodbreak| macro is an abbreviation for `|\par\penalty-500|'.
This is a good thing to insert in your manuscript when proofreading, if
you are willing to stretch some page a little bit extra in order to improve the
following one. Later on if you make another change so that this |\goodbreak|
command does not appear near the bottom of a page, it will have no effect;
thus it is not as drastic as |\eject|.

\danger The most interesting macro that plain \TeX\ provides for page make-up
is called ^|\filbreak|. It means, roughly, ``Break the page here and fill the
bottom with blank space, unless there is room for more
copy that is itself followed by |\filbreak|.'' Thus if you put |\filbreak|
at the end of every paragraph, and if your paragraphs aren't too long,
every page break will occur between paragraphs, and \TeX\ will fit as many
paragraphs as possible on each page. The precise meaning of\/ |\filbreak|~is
\begintt
\vfil\penalty-200\vfilneg
\endtt
according to Appendix B\null; and this simple combination of \TeX's primitives
produces the desired result: If a break is taken at the |\penalty-200|,
the preceding |\vfil| will fill the bottom of the page with blank space,
and the ^|\vfilneg| will be discarded after the break; but if no break
is taken at the penalty, the |\vfil| and |\vfilneg| will cancel each other
and have no effect.

\danger Plain \TeX\ also provides a ^|\raggedbottom| command, which is a
vertical analog of\/ ^|\raggedright|: It tells \TeX\ to permit a small
amount of variability in the bottom margins on different pages, in order
to make the other spacing uniform.

\ddanger We saw in Chapter 14 that breakpoints for paragraphs are chosen
by computing ``demerits'' for each line and summing them over all lines.
The situation for pages is simpler because each page is considered separately.
\TeX\ figures the ``^{cost}'' of a page break by using the following formula:
\begindisplay
$\displaystyle{c=\cases{
  p,&if $b<\infty$ and $p\le-10000$ and $q<10000$;\cr
  b+p+q,&if $b<10000$ and $-10000<p<10000$ and $q<10000$;\cr
  100000,&if $b=10000$ and $-10000<p<10000$ and $q<10000$;\cr
  \infty,&if ($b=\infty$ or $q\ge10000$) and $p<10000$.\cr}}$
\enddisplay
Here $b$ is the ^{badness} of the page that would be formed if a break
were chosen here; $p$~is the penalty associated with the current
breakpoint; and $q$~is `^|\insertpenalties|', the sum of all penalties for
split insertions on the page, as explained below. Vertical badness is
computed by the same rules as horizontal badness; it is an integer between
0 and~10000, inclusive, except when the box is overfull, when it
is~$\infty$ (infinity).
^^{infinite badness}

\ddanger When a page is completed, it is removed from the main vertical
list and passed to an ``^{output routine},'' as we will see later; so its
boxes and glue eventually disappear from \TeX's memory. The remainder of
the main vertical list exists in two parts:  First comes the ``^{current
page},'' which contains all the material that \TeX\ has considered so far
as a candidate for the next page to be broken off; then there are
``^{recent contributions},'' i.e., items that will be moved to the current
page as soon as \TeX\ finds it convenient to do so. If you say
^|\showlists|, \TeX\ will display the contents of the current page and the
recent contributions, if any, on your log file. \ (The example in
Chapter~13 doesn't show any such lists because they were both empty in
that case. Chapter~24 explains more about \TeX's timing.)

\ddanger Whenever \TeX\ is moving an item from the top of the ``recent
contributions'' to the bottom of the ``current page,'' it discards a
^{discardable item} (glue, kern, or penalty) if the current page does not
contain any boxes. This is how glue disappears at a page break. Otherwise
if a discardable item is a legitimate breakpoint, \TeX\ calculates the
cost~$c$ of breaking at this point, using the formula that we have just
discussed. If the resulting~$c$ is less than or equal to the smallest cost
seen so far on the current page, \TeX\ remembers the current
breakpoint as the best so far. And if $c=\infty$ or if $p\le-10000$, \TeX\
seizes the initiative and breaks the page at the best remembered
breakpoint.  Any material on the current page following that best
breakpoint is moved back onto the list of recent contributions, where it
will be considered again; thus the ``current page'' typically gets more
than one page's worth of material before the breakpoint is chosen.

\ddanger This procedure may seem mysterious until you see it in action.
Fortunately, there is a convenient way to watch it;
you can set ^|\tracingpages||=1|, thereby instructing \TeX\ to put its
page-cost calculations into your log file. For example, here is what
appeared on the log file when the author used |\tracingpages=1| at the
beginning of the present chapter:
\begintt
%% goal height=528.0, max depth=2.2
% t=10.0 g=528.0 b=10000 p=150 c=100000#
% t=22.0 g=528.0 b=10000 p=0 c=100000#
% t=34.0 g=528.0 b=10000 p=0 c=100000#
|kern-3pt
|qquad|hbox|bgroup|rm|vdots|quad(25 similar lines are being omitted here)|egroup
% t=346.0 plus 2.0 g=528.0 b=10000 p=0 c=100000#
% t=358.0 plus 2.0 g=528.0 b=10000 p=150 c=100000#
% t=370.02223 plus 2.0 g=528.0 b=10000 p=-100 c=100000#
% t=398.0 plus 5.0 minus 2.0 g=528.0 b=10000 p=0 c=100000#
% t=409.0 plus 5.0 minus 2.0 g=528.0 b=10000 p=0 c=100000#
% t=420.0 plus 5.0 minus 2.0 g=528.0 b=10000 p=150 c=100000#
% t=431.0 plus 5.0 minus 2.0 g=528.0 b=10000 p=-100 c=100000#
% t=459.0 plus 8.0 minus 4.0 g=528.0 b=10000 p=0 c=100000#
% t=470.0 plus 8.0 minus 4.0 g=528.0 b=10000 p=0 c=100000#
% t=481.0 plus 8.0 minus 4.0 g=528.0 b=10000 p=0 c=100000#
% t=492.0 plus 8.0 minus 4.0 g=528.0 b=10000 p=0 c=100000#
% t=503.0 plus 8.0 minus 4.0 g=528.0 b=3049 p=0 c=3049#
% t=514.0 plus 8.0 minus 4.0 g=528.0 b=533 p=150 c=683#
% t=525.0 plus 8.0 minus 4.0 g=528.0 b=5 p=-100 c=-95#
% t=553.0 plus 11.0 minus 6.0 g=528.0 b=* p=0 c=*
\endtt
This trace output is admittedly not ``user-friendly'' in appearance, but
after all it comes from deep inside \TeX's bowels where things have been
reduced to numeric calculations. You can learn to read it with a little
practice, but you won't need to do so very often unless you need to
plunge into page-breaking for special applications. Here's what it
means: The first line, which starts with `|%%|', ^^{percent percent}
is written when the first box or insertion enters the current page list;
it shows the ``^{goal height}'' and the ``^{max depth}'' that will be
used for that page (namely, the current values of\/ ^|\vsize| and
^|\maxdepth|). In the present manual we have |\vsize=44pc| and
|\maxdepth=2.2pt|; dimensions in the log file are always displayed in
points. The subsequent lines, which start with a single `|%|', ^^{percent}
are written whenever a legal breakpoint is being moved from the list of
recent contributions to the current page list. Every |%|~line shows $t$,
which is the total height so far if a page break were to occur,
and $g$, which is the goal height; in this example $g$~stays fixed at
$528\pt$, but $g$ would have decreased if insertions such as footnotes had
occurred on the page. The values of~$t$ are steadily increasing from 10 to
22 to~34, etc.; baselines are $12\pt$ apart at the top of the page and
$11\pt$ apart at the bottom (where material is set in nine-point type).
We are essentially seeing one |%|~line per hbox of text being placed on
the current page. However, the |%|~lines are generated by the penalty or
glue items that follow the hboxes, not by the boxes themselves. Each
|%|~line shows also the badness~$b$, the penalty~$p$, and the cost~$c$
associated with a breakpoint; if this cost is the best so far, it is
marked with a `|#|' sign, ^^{sharp} meaning that ``this breakpoint will be
used for the current page if nothing better comes along.'' Notice that the
first 40 or so breaks all have $b=10000$, since they are so bad that \TeX\
considers them indistinguishable; in such cases $c=100000$, so \TeX\ simply
accumulates material until the page is full enough to
have $b<10000$. A penalty of 150 reflects the ^|\clubpenalty| or the
^|\widowpenalty| that was inserted as described in Chapter~14.  The three
lines that say |p=-100| are the breakpoints between ``dangerous bend''
paragraphs; these came from ^|\medbreak| commands.  The notation |b=|^|*| and
|c=*| on the final line means that $b$ and~$c$ are infinite; the total
height of $553\pt$ cannot be reduced to $528\pt$ by shrinking the
available glue. Therefore the page is ejected at the best previous place,
which turns out to be a pretty good break: |b=5| and |p=-100| yield a
net cost of $-95$.

\ddangerexercise Suppose the paragraph at the bottom of the example page
had been one line shorter; what page break would have been chosen?
\answer The last three page-break calculations would have been
\begintt
% t=503.0 plus 8.0 minus 4.0 g=528.0 b=3049 p=150 c=3199#
% t=514.0 plus 8.0 minus 4.0 g=528.0 b=533 p=-100 c=433#
% t=542.0 plus 11.0 minus 6.0 g=528.0 b=* p=0 c=*
\endtt
so the break would have occurred at the same place. The badness would have
been~533, but the page would still have looked tolerable. \ (On the other
hand if that paragraph had been two lines shorter instead of one,
the first two lines of the next ``dangerous bend'' paragraph
would have appeared on that page; the natural height $t=531\pt$ would have
been able to shrink to $g=528\pt$ because the three ``medskips'' on
the page would have had a total shrinkability of $6\pt$. This would certainly
have been preferable to a stretched-out page whose badness was~3049; but the
author might have seen it and written another sentence or two, so that
the paragraph would not have been broken up. After all, this manual is supposed
to be an example of good practice.)

\ddangerexercise The last two ``\kern.5pt|%| lines'' of this example show
the natural height of~$t$ jumping by $28\pt$, from 525.0 to~553.0. Explain
why there was such a big jump.
\answer The next legal break after the beginning of a dangerous bend
paragraph occurs $28\pt$ later, because there is $6\pt$ additional space for
a |\medskip|, followed by two lines of $11\pt$ each. \TeX\ does not
allow breaking between those two lines; the ^|\clubpenalty| is set briefly
to 10000 in Appendix~E\null, since the dangerous bend symbol is two lines tall.

\ddanger The ^|\maxdepth| parameter tells \TeX\ to raise the bottom box
on the page if that box has too much depth, so that the depth of the
constructed page will not exceed a specified value. \ (See the discussion of
^|\boxmaxdepth| in Chapter~12.) \ In our example |\maxdepth=2.2pt|, and
the influence of this parameter can be seen in the line that says
`|% t=370.02223|'. Ordinarily $t$~would have been 370.0 at that breakpoint;
but the hbox preceding it was unusual because it contained the letter
|j| in ^|\tt|, and a 10-point typewriter-style |j| descends $2.22223\pt$
below the baseline.  Therefore \TeX\ figured badness as if the hbox were
$.02223\pt$ higher and only $2.2\pt$ deep.

\ddanger Notice that the first ``\kern.5pt|%| line'' of our example says
|t=10.0|; this is a consequence of another parameter, called ^|\topskip|.
Glue disappears at a page break, but it is desirable to produce pages
whose top and bottom baselines occur in predetermined positions, whenever
possible; therefore \TeX\ inserts special glue just before the first box
on each page.  This special glue is equal to |\topskip|, except that the
natural space has been decreased by the height of the first box, or it has
been set to zero in lieu of a negative value. For example, if\/
|\topskip=20pt plus2pt|, and if the first box on the current page is
$13\pt$ tall, \TeX\ inserts `|\vskip7pt plus2pt|' just above that box.
Furthermore, if the first box is more than $20\pt$ tall, `|\vskip0pt
plus2pt|' is inserted. But this example is atypical, since the |\topskip|
glue usually has no stretchability or shrinkability; plain \TeX\ sets
|\topskip=10pt|.

\ddangerexercise Assume that |\vsize=528pt|, |\maxdepth=2.2pt|,
|\topskip=10pt|, and that no |\insert| commands are being used. \TeX\ will
make pages that are $528\pt$ high, and the following two statements will
normally be true: (a)~The baseline of the topmost box on the page will be
$10\pt$ from the top, i.e., $518\pt$ above the baseline of the page itself.
(b)~The baseline of the bottommost box on the page will coincide with the
baseline of the page itself. Explain under what circumstances (a) and~(b)
will fail.
\answer A page always contains at least one box, if there are no
insertions, since the legal breakpoints are discarded otherwise.
Statement~(a) fails if the height of the topmost box exceeds $10\pt$.
Statement~(b) fails if the depth of the bottommost box exceeds $2.2\pt$, or
if some glue or kern comes between the bottommost box and the page break
(unless that glue or kern exactly cancels the depth of the box).

\ddanger Since |\vsize|, |\maxdepth|, and |\topskip| are
parameters, you can change them at any time; what happens if you do?
Well, \TeX\ salts away the values of\/ |\vsize| and |\maxdepth| when
it prints the ``\kern.5pt|%%|~line,'' i.e., when the first box or insertion
occurs on the current page; subsequent changes to those two parameters
have no effect until the next current page is started. On the other hand,
\TeX\ looks at |\topskip| only when the first box is being contributed
to the current page. If insertions occur before the first box, the
|\topskip| glue before that box is considered to be a valid breakpoint;
this is the only case in which a completed page might not contain a box.

\ddanger You can look at the $t$ and $g$ values that are used in page
breaking by referring to the \<dimen> values `^|\pagetotal|' and
`^|\pagegoal|', respectively.  You can even change them (but let's hope
that you know what you are doing).  For example, the command
|\pagegoal=500pt| overrides the previously saved value of\/ |\vsize|.
Besides |\pagetotal|, which represents the accumulated natural height,
\TeX\ maintains the quantities ^|\pagestretch|, ^|\pagefilstretch|,
^|\pagefillstretch|, ^|\pagefilllstretch|, ^|\pageshrink|, and
^|\pagedepth|.  When the current page contains no boxes, |\pagetotal| and
its relatives are zero and |\pagegoal| is $16383.99998\pt$ (\TeX's largest
\<dimen>); changing their values has no effect at such times.  The integer
$q$ in the formula for page costs is also available for inspection and
change; it is called ^|\insertpenalties|.

\ddanger Page breaking differs from line breaking in one small respect
that deserves mention here: If you say |\eject\eject|, the second |\eject|
is ignored, because it is equivalent to |\penalty-10000| and penalties are
discarded after a page break. But if you say |\break\break| in a
paragraph, the second ^|\break| causes an empty line, because penalties
are discarded after a break in a paragraph only if they do not belong to
the final sequence of breakpoints. This technicality is unimportant in
practice, because |\break\break| isn't a good way to make an empty line;
that line will usually be an underfull hbox, since it has only the
|\leftskip| and |\rightskip| glue in it.  Similarly, `|\eject\eject|'
would not be a good way to make an empty page, even if \TeX\ were to
change its rules somehow so that an |\eject| would never be ignored. The
best way to eject an ^{empty page} is to say `|\eject\line{}\vfil\eject|',
and the best way to create an ^{empty line} is
`|\break\hbox{}\hfil\break|'. Both of these avoid underfull boxes.

\danger You are probably wondering how page numbers and such things
get attached to pages. The answer is that \TeX\ allows you to do further
processing after each page break has been chosen; a special ``output
routine'' goes into action before pages actually receive their final
form. Chapter~23 explains how to construct output routines and how to
modify the output routine of plain~\TeX.

\danger Every once in a~while, \TeX\ will produce a really awful-looking
page and you will wonder what happened. For example, you might get just
one paragraph and a lot of white space, when some of the text on the
following page would easily fit into the white space. The reason for such
apparently anomalous behavior is almost always that no good page break
is possible; even the alternative that looks better to you is quite terrible
as far as \TeX\ is concerned! \TeX\ does not distinguish between two
choices that both have 10000 units of badness or more, even though some bad
breaks do look much worse than others. The solution in such cases is to
insert |\eject| or |\vfill\eject| in some acceptable spot, or to
revise the manuscript. If this problem arises frequently, however,
you probably are using a format that sets overly strict limitations
on page format; try looking at the output of\/ |\tracingpages| and
modifying some of \TeX's parameters, until you have better luck.

\danger The remainder of this chapter is about insertions: things like
footnotes and ^{illustrations}, and how they interact with page breaks.
Before we discuss the primitive operations by which \TeX\ deals with
insertions, we will take a look at the facilities that plain \TeX\ provides
at a higher level.

\danger Illustrations can be inserted in several ways using plain \TeX.
The simplest of these is called a ``^{floating topinsert}''; you say
\begindisplay
^|\topinsert|\<vertical mode material>^|\endinsert|
\enddisplay
and \TeX\ will attempt to put the vertical mode material at the top of the
current page.  If there's no room for such an insertion on this page, \TeX\
will insert it at the top of the next page. The \<vertical mode material>
can contain embedded paragraphs that temporarily interrupt vertical mode
in the usual way; for example:
\begintt
\topinsert \vskip 2in
\hsize=3in \raggedright
\noindent{\bf Figure 3.} This is the caption to the
third illustration of my paper. I have left two inches
of space above the caption so that there will be room
to introduce special artwork. \endinsert
\endtt
The ^{caption} in this example will be set ^{ragged-right} in a 3-inch
column at the left of the page. Plain \TeX\ automatically adds a
``^{bigskip}'' below each topinsert; this will separate the caption from
the text. The effects of\/ |\hsize=3in| and |\raggedright| do not extend
past the |\endinsert|, since ^{grouping} is implied.

\dangerexercise Modify this example so that the caption is moved over next to
the right margin, instead of appearing at the left.
\answer |\topinsert\vskip2in\rightline{\vbox{\hsize|\stretch|...|\stretch
|artwork.}}\endinsert|
does the job. But it's slightly more efficient to avoid ^|\rightline| by
changing ^|\leftskip| as follows:
`|\leftskip=\hsize \advance\leftskip by-3in|'.
Then \TeX\ doesn't have to read the text of the caption twice.

\danger Similarly, if you say `^|\pageinsert|\stretch\<vertical mode
material>\stretch |\endinsert|', the vertical mode material will be
justified to the size of a full page (without a bigskip below it);
the result will appear on the following page.

\danger There's also `^|\midinsert|\stretch\<vertical mode material>\stretch
|\endinsert|', which tries first to insert the material in place, wherever
you happen to be, in the middle of the current page. If there is enough
room, you get the effect of^^|\bigskip|^^|\bigbreak|
\begindisplay
|\bigskip\vbox{|\<vertical mode material>|}\bigbreak|
\enddisplay
otherwise the |\midinsert| is effectively converted to a |\topinsert|.
There is a slight probability that |\midinsert| will not find the
best placement, because \TeX\ is sometimes processing text ahead of the
current page. You may want to say `^|\goodbreak|' just before |\midinsert|.

\danger You should use the commands |\topinsert|, |\pageinsert|,
|\midinsert| in vertical mode (i.e., between paragraphs), not inside of
boxes or other insertions.

\danger If you have two or more |\topinsert| or |\pageinsert| commands in
quick succession, \TeX\ may need to carry them over to several subsequent
pages; but they will retain their relative order when they are
carried over.  For example, suppose you have pages that are nine inches
tall, and suppose you have already specified 4~inches of text for some
page, say page~25.  Then suppose you make seven topinserts in a row, of
respective sizes $1,2,3,9,3,2,1$ inches; the 9-inch one is actually a
|\pageinsert|. What happens?  Well, the first and second will appear at
the top of page 25, followed by the 4~inches of copy you have already
typed; that copy will immediately be followed by two more inches that you
type after the seven inserts.  The third topinsert will appear at the top
of page~26, followed by six more inches of text; the fourth will fill
page~27; and the remaining three will appear at the top of page~28.

\dangerexercise What would happen in the example just discussed if the final
1-inch insertion were a |\midinsert| instead of a |\topinsert|?
\answer It would appear on page~25, since it does fit there. A |\midinsert|
will jump ahead of other insertions only if it is not carried over to another
page; for example, if the second 3-inch insertion were a |\midinsert|, it would
not appear on page~26, because it is converted to a |\topinsert| as soon as the
|\midinsert| macro notices that the insertion is too big for page~25.

\danger At the end of a paper, you probably want to make sure that no
insertions are lost; and at the end of a chapter, you probably want to make
sure that no insertions float into the following chapter. Plain \TeX\ will
flush out all remaining insertions, with blank space filling the bottom of
incomplete pages, if you say `|\vfill|^|\supereject|'.

\danger Besides illustrations that are inserted at the top of a page,
plain \TeX\ will also insert ^{footnotes} at the bottom of a page. The
^|\footnote| macro is provided for use within paragraphs;\footnote*{Like
this.} for example, the footnote in the present sentence was typed
in the following way:
\begintt
... paragraphs;\footnote*{Like this.} for example, ...
\endtt
There are two parameters to a |\footnote|; first comes the ^{reference mark},
which will appear both in the paragraph\footnote{**}{The author typed
`|paragraph\footnote{**}{The author ...}|' here.} and in the footnote itself,
and then comes the text of the footnote.\footnote{$^{45}$}{And
`|footnote.\footnote{$^{45}$}{And ...}|' here. The footnotes in this manual
appear in smaller type, and they are set with hanging indentation;
furthermore a smallskip occurs between footnotes on the same page. But
in plain \TeX, footnotes are typeset with the normal size of type, with
|\textindent| used for the reference mark, and without extra smallskips.
The |\textindent| macro is like |\item|, but it omits hanging
indentation.}
The latter text may be several paragraphs long, and it may contain
displayed equations and such things, but it should not involve other
^^|\textindent| ^^|\item| % these are in a split footnote, after the split!
insertions. \TeX\ will ensure that each footnote occurs at the bottom of
the same page as its reference.\footnote\dag{Printers often use the
symbols |\dag| (\dag), |\ddag| (\ddag), |\S|~(\S), and |\P|~(\P) as
reference marks; sometimes also |$\|\||$| ($\Vert$).
You can say, e.g., `|\footnote\dag{...}|'.} A long
footnote will be split, if necessary, and continued at the bottom of the
following page, as you can see in the ^^|\dag|^^|\ddag|^^|\S|^^|\P|^^|\Vert|
somewhat contrived example that appears here. Authors who are interested
in good exposition should avoid footnotes whenever possible, since
footnotes tend to be ^^{Gibbon} distracting.\footnote\ddag{Yet Gibbon's
{\sl Decline and Fall\/} would not have been the same without footnotes.}

\danger The |\footnote| macro should be used only in paragraphs or hboxes
that are contributed to \TeX's main vertical list; insertions will be lost
if they occur inside of boxes that are inside of boxes. Thus, for example,
you should not try to put a |\footnote| into a subformula of a math formula.
But it's OK to use footnotes within ^|\centerline|, e.g., ^^{Thor}
\begintt
\centerline{A paper by A. U. Thor%
  \footnote*{Supported by NSF.}}
\endtt
or even on the outer level of a table entry inside an ^|\halign|.

\ddanger Topinserts work fine by themselves, and footnotes work fine by
themselves, but complications can arise when you try to mix them in
devious ways. For example, if a |\pageinsert| floats to the page that
follows a long footnote that had to be broken, both of the held-over
insertions may try to force themselves onto the same page, and an overfull
vbox may result. Furthermore, insertions cannot appear within insertions,
so you can't use |\footnote| within a |\topinsert|.  If you really need a
footnote in some caption, there's a ^|\vfootnote| macro that can be used
in vertical mode. To use it, you put a reference mark like~`|*|' in the
caption, and then you say `|\vfootnote*{The footnote}|' somewhere on the
page where you guess that the caption will finally fall.  In such complex
circumstances you might want to rethink whether or not you are really
using the most appropriate format for the exposition of your ideas.

\ddanger Chapter 24 explains the exact rules about ^{migration} of
vertical-mode material (like footnotes) from horizontal lists to the
enclosing vertical list. Insertions, marks, and the results of\/ ^|\vadjust|
all migrate in the same fashion.

\danger Now let's study the primitives of \TeX\ that are used to construct
macros like |\topinsert| and |\footnote|. We are about to enter behind the
scenes into a sublanguage of \TeX\ that permits users to do complex
manipulations with boxes and glue. Our discussion will be in two parts:
First we shall consider \TeX's ``^{registers},'' with which a user can
do ^{arithmetic} related to typesetting; and then we shall discuss the insertion
items that can appear in horizontal and vertical lists. Our discussion of the
first topic (registers) will be marked with single dangerous-bend signs,
since registers are of general use in advanced applications of \TeX,
whether or not they relate to insertions. But the second topic will be marked
with double dangerous-bend signs, since insertions are rather esoteric.

\danger \TeX\ has 256 registers called ^|\count||0| to |\count255|, each capable
of containing integers between $-2147483647$ and $+2147483647$, inclusive;
^^\<number>
i.e., the magnitudes should be less than $2^{31}$. \TeX\ also has 256 registers
called ^|\dimen||0| to |\dimen255|, each capable of containing a ^\<dimen>
(see Chapter~10). There are another 256 registers called ^|\skip||0| to
|\skip255|, each containing ^\<glue> (see Chapter~12); and ^|\muskip||0| to
|\muskip255|, each containing ^\<muglue> (see Chapter~18). You can assign
new values to these registers by saying
\begindisplay
|\count|\<number> |=| \<number>\cr
|\dimen|\<number> |=| \<dimen>\cr
|\skip|\<number> |=| \<glue>\cr
|\muskip|\<number> |=| \<muglue>\cr
\enddisplay
and then you can add or subtract values of the same type by saying^^|\advance|
\begindisplay
|\advance\count|\<number> |by| \<number>\cr
|\advance\dimen|\<number> |by| \<dimen>\cr
|\advance\skip|\<number> |by| \<glue>\cr
|\advance\muskip|\<number> |by| \<muglue>\cr
\enddisplay
For example, `|\dimen8=\hsize \advance\dimen8 by 1in|' sets register
|\dimen8| to an inch more than the current value of the normal line size.

\danger If infinite glue components are added, lower order infinities disappear.
For example, after the two commands
\begintt
\skip2 = 0pt plus 2fill minus 3fill
\advance\skip2 by 4pt plus 1fil minus 2filll
\endtt
the value of\/ |\skip2| will be $4\pt$ plus $2\,{\rm fill}$ minus
$2\,{\rm filll}$.

\danger Multiplication and division are possible too, but only by integers.
For example, `^|\multiply||\dimen4 by 3|' triples the value of\/ |\dimen4|, and
`^|\divide||\skip5 by 2|' cuts in half all three components of the glue that is
currently registered in |\skip5|.
You shouldn't divide by zero, nor should you multiply by numbers that
will make the results exceed the register capacities. Division of a positive
integer by a positive integer discards the remainder, and the sign of the
result changes if you change the sign of either operand. For example,
14~divided by~3 yields~4; $-14$~divided by~3 yields~$-4$;
$-14$~divided by~$-3$ yields~4. Dimension values are integer multiples
of~^{sp} (scaled points).

\danger You can use any |\count| register in the context of a ^\<number>,
any |\dimen| register in the context of a ^\<dimen>, any |\skip| register
in the context of ^\<glue>, and any |\muskip| register in the context of
^\<muglue>. For example, `|\hskip\skip1|' puts horizontal glue into a
list, using the value of\/ |\skip1|; and if\/ |\count5| is 20,
the~command `|\advance\dimen20 by\dimen\count5|' is equivalent to
`|\multiply\dimen20 by 2|'.

\danger A |\dimen| register can be used also in the context of a \<number>,
and a |\skip| register can be used as a \<dimen> or a \<number>.
\TeX\ converts \<glue> to \<dimen> by omitting the stretch and shrink
components, and it converts \<dimen> to \<number> by assuming units
of~sp (scaled points). For example, if\/ |\skip1| holds the value
$1\pt$ plus~$2\pt$, then `|\dimen1=\skip1|' sets |\dimen1| equal
to~$1\pt$; and the commands `|\count2=\dimen1|' or `|\count2=\skip1|' will
set |\count2| equal to~65536. These rules also apply to \TeX's internal
parameters; for example, `|\dimen2=\baselineskip|' will set |\dimen2| to
the natural space component of the current baselineskip glue.

\dangerexercise Test your knowledge of \TeX's registers by stating the
results of each of the following commands when they are performed in sequence:
\begintt
\count1=50   \dimen2=\count1pt   \divide\count1 by 8
\skip2=-10pt plus\count1fil minus\dimen2
\multiply\skip2 by-\count1  \divide\skip2 by \dimen2  \count6=\skip2
\skip1=.5\dimen2 plus\skip2 minus\count\count1fill
\multiply\skip2 by\skip1   \advance\skip1 by-\skip2
\endtt
\answer Set |\count1| to 50,
then |\dimen2| to~$50\pt$,
then |\count1| to~6,
then |\skip2| to~$-10\pt$ plus~$6\,{\rm fil}$ minus~$50\pt$,
then |\skip2| to~$60\pt$ plus~$-36\,{\rm fil}$ minus~$-300\pt$,
then |\skip2| to~$1\,{\rm sp}$ minus~$-6\,{\rm sp}$,
then |\count6| to~1,
then |\skip1| to~$25\pt$ plus~$1\,{\rm sp}$ minus~$1\,{\rm fill}$,
then |\skip2| to~$25\pt$ minus~$-150\pt$,
then |\skip1| to~$0\pt$ plus~$1\,{\rm sp}$ minus~$1\,{\rm fill}$.

\dangerexercise What is in |\skip5| after the following three commands have
acted?
\begintt
\skip5=0pt plus 1pt
\advance\skip5 by \skip4  \advance\skip5 by -\skip4
\endtt
\answer If\/ |\skip4| has infinite stretchability, |\skip5| will be zero;
otherwise it will be $0\pt$ plus~$1\pt$.

\dangerexercise (For mathematicians.) Explain how to ^{round} |\dimen2|
to the nearest multiple of\/ |\dimen3|, assuming that |\dimen3| is positive.
\answer |\advance\dimen2 by\ifnum\dimen2<0 -\fi.5\dimen3|\parbreak
|\divide\dimen2 by\dimen3 \multiply\dimen2 by\dimen3|

\danger The registers obey \TeX's ^{group structure}. For example,
changes to |\count3| inside |{...}| will not affect the value of
|\count3| outside. Therefore \TeX\ effectively has more than 256
registers of each type. If you want the effect of a register command
to transcend its group, you must say ^|\global| when you change the value.

\dangerexercise What is in |\count1| after the following sequence of commands?
\begintt
\count1=5 {\count1=2 \global\advance\count1by\count1
  \advance\count1by\count1}
\endtt
\answer |\count1| takes the values 5, then~2 (the old 5 is saved),
then~4 (which is made global), then~8 (and 4~is saved); finally the value~4 is
restored, and that is the answer. \ (For further remarks, see the discussion
of\/ |\tracingrestores| in Chapter~27.)

\danger The first ten |\count| registers, |\count0| through |\count9|,
are reserved for a special purpose:
\TeX\ displays these ten counts on your terminal whenever outputting
a page, and it transmits them to the output file as an identification
of that page. The counts are separated by decimal points on your terminal,
with trailing `|.0|' patterns suppressed. Thus, for example, if
|\count0=5| and |\count2=7| when a page is being shipped out to the
^|dvi| file, and if the other count registers are zero, \TeX\ will
type `|[5.0.7]|'. Plain \TeX\ uses |\count0| for the page number, and it
keeps |\count1| through~|\count9| equal to zero; that is why you see just
`^|[1]|' when page~1 is being output. In more complex applications the
page numbers can have further structure; ten counts are shipped out so
that there will be plenty of identification.

\danger It's usually desirable to have symbolic names for registers.
\TeX\ provides a ^|\countdef| command (similar to |\chardef|,
cf.~Chapter~8), which makes it easy to do this: You just say
\begintt
\countdef\chapno=28
\endtt
and |\chapno| is henceforth an abbreviation for |\count28|. Similar
commands ^|\dimendef|, ^|\skipdef|, and ^|\muskipdef| are available
for the other types of numeric registers. After a control sequence
has been defined by |\countdef|, it can be used in \TeX\ commands exactly
as if it were an integer ^{parameter} like |\tolerance|. Similarly,
|\dimendef| effectively creates a new dimension parameter, |\skipdef|
effectively creates a new glue parameter, and |\muskipdef| effectively
creates a new muglue parameter.

\danger Besides the numerical registers, \TeX\ also has 256 box registers
called ^|\box||0| to |\box255|. A~box register gets a value when you say
^|\setbox|\<number>|=|^\<box>; for example, `|\setbox3=\hbox{A}|' sets
|\box3| to an hbox that contains the single letter~|A|. Several other
examples of\/ |\setbox| have already appeared in Chapter~12. Chapter~10
points out that `|2\wd3|' is a \<dimen> that represents twice the width
of\/ |\box3|; similarly, ^|\ht|\<number> and ^|\dp|\<number> can be used to
refer to the height and depth of a given box register. ^^|\wd|

\danger Box registers are local to groups just as arithmetic registers are.
But there's a big difference between box registers and all the rest: When
you use a |\box|, it loses its value. For example, the construction
`|\raise2pt\box3|' in a horizontal list not only puts the contents of
|\box3| into the list after raising it by~$2\pt$, it also makes |\box3|
void. \TeX\ does this for efficiency, since it is desirable to avoid copying
the contents of potentially large boxes. If you want to use a box register
without wiping out its contents, just say `^|\copy|' instead of `|\box|';
for example, `|\raise2pt\copy3|'.

\danger Another way to use a box register is to extract the inside of
an hbox by saying `^|\unhbox|'. This annihilates the contents of the
register, like `|\box|' does, and it also removes one level of boxing. For
example, the commands
\begintt
\setbox3=\hbox{A} \setbox3=\hbox{\box3 B}
\setbox4=\hbox{A} \setbox4=\hbox{\unhbox4 B}
\endtt
put |\hbox{\hbox{A}B}| into |\box3| and |\hbox{AB}| into |\box4|.
Similarly, ^|\unvbox| unwraps a vbox.  If you want to construct a large
box by accretion (e.g., a ^{table of contents}), it is best to use
|\unhbox| or |\unvbox| as in the |\setbox4| example; otherwise you use
more of \TeX's memory space, and you might even obtain boxes inside boxes
nested to such a deep level that hardware or software limits are exceeded.

\danger The operations ^|\unhcopy| and ^|\unvcopy| are related to
|\unhbox| and |\unvbox| as |\copy| is to |\box|. \ (But their
names are admittedly peculiar.)

\danger An unboxing operation ``unsets'' any glue that was set at the box's
outer level. For example, consider the sequence of commands
\begintt
\setbox5=\hbox{A \hbox{B C}} \setbox6=\hbox to 1.05\wd5{\unhcopy5}
\endtt
This makes |\box6| five percent wider than |\box5|;
the glue between |A| and |\hbox{B C}| stretches to
make the difference, but the glue inside the inner hbox does not change.

\danger A box register is either ``^{void}'' or it contains an hbox or a vbox.
There is a difference between a void register and one that contains an
empty box whose height, width, and depth are zero; for example, if\/ |\box3|
is void, you can say |\unhbox3| or |\unvbox3| or |\unhcopy3| or |\unvcopy3|,
but if\/ |\box3| is equal to |\hbox{}| you can say only |\unhbox3| or
|\unhcopy3|. If you say `|\global\setbox3=|\<box>', register |\box3| will
become ``globally void'' when it is subsequently used or unboxed.

\dangerexercise What is in register |\box5| after the following commands?
\begintt
\setbox5=\hbox{A}  \setbox5=\hbox{\copy5\unhbox5\box5\unhcopy5}
\endtt
\answer |\hbox{\hbox{A}A}|. After `|\unhbox5|', |\box5| is void; |\unhcopy5|
yields nothing.

\dangerexercise And what's in |\box3| after ^^{grouping with box registers}
`|{\global\setbox3=\hbox{A}\setbox3=\hbox{}}|'?\kern-1pt
\answer |\hbox{A}|. But after `|{\global\setbox3=\hbox{A}\setbox3=\box3}|',
|\box3| will be void.

\danger If you are unsure about how \TeX\ operates on its registers, you can
experiment online by using certain `|\show|' commands. For example,
\begintt
\showthe\count1  \showthe\dimen2  \showthe\skip3
\endtt
will display the contents of\/ |\count1|, |\dimen2|, and |\skip3|; and
`^|\showbox||4|' will display the contents of\/ |\box4|. ^^|\showthe|
Box contents will appear only in the log file, unless you say
`|\tracingonline=1|'. Plain \TeX\ provides a macro `^|\tracingall|'
that turns on every possible mode of interaction, including
^|\tracingonline|. The author used these features to check the answers to
several of the exercises above.

\danger Large applications of \TeX\ make use of different sets of macros
written by different groups of people. Chaos would reign if a register
like |\count100|, say, were being used simultaneously for different purposes
in different macros. Therefore plain \TeX\ provides an ^{allocation} facility;
cooperation will replace confusion if each ^{macro writer} uses these
conventions. The idea is to say, e.g., `^|\newcount|' when you want
to dedicate a |\count| register to a special purpose. For example, the
author designed a macro called `|\exercise|' to format the exercises
in this manual, and one of the features of\/ |\exercise| is that it
computes the number of the current exercise. The format
macros in Appendix~E reserve a |\count| register for this purpose by saying
\begintt
\newcount\exno
\endtt
and then the command `|\exno=0|'
is used at the beginning of each chapter. Similarly, `|\advance\exno by1|'
is used whenever a new exercise comes along, and `|\the\exno|'
is used to typeset the current exercise number. The |\newcount| operation
assigns a unique count register to its argument |\exno|, and it defines
|\exno| with a ^|\countdef| command. All of the other format macros
are written without the knowledge of exactly which |\count| register
actually corresponds to |\exno|.

\danger Besides |\newcount|, plain \TeX\ provides ^|\newdimen|,
^|\newskip|, ^|\newmuskip|, and ^|\newbox|; there also are ^|\newtoks|,
^|\newread|, ^|\newwrite|, ^|\newfam|, and ^|\newinsert|, for
features we haven't discussed yet. Appendices~B and~E contain
several examples of the proper use of allocation. In the cases of
|\newbox|, |\newread|, etc., the allocated number is defined by |\chardef|.
For example, if the command `|\newbox\abstract|' is used to define a
box register that will contain an abstract, and if the |\newbox| operation
decides to allocate |\box45| for this purpose, then it defines the
meaning of\/ |\abstract| by saying `|\chardef\abstract=45|'. \TeX\ allows
^|\chardef|'d quantities to be used as integers, so that you can say
|\box\abstract| and |\copy\abstract|, etc. \ (There is no |\boxdef|
command.)

\newcount\notenumber
\def\clearnotenumber{\notenumber=0\relax}
\def\note{\advance\notenumber by1 \footnote{$^{\the\notenumber}$}}
\clearnotenumber
\dangerexercise Design a |\note| macro that produces footnotes numbered
sequentially. For example,\note{First note.} it should produce
the footnotes here\note{Second note.} if you type
\begintt
... example,\note{First note.} it should produce
the footnotes here\note{Second note.} if ...
\endtt
(Use |\newcount| to allocate a |\count| register for the footnotes.)
\answer |\newcount\notenumber|\parbreak
|\def\clearnotenumber{\notenumber=0\relax}|\parbreak
|\def\note{\advance\notenumber by 1|\parbreak
|  \footnote{$^{\the\notenumber}$}}|

\danger Sometimes, however, you want to use a register just for temporary
storage, and you know that it won't conflict with anybody else's macros.
Registers |\count255|, |\dimen255|, |\skip255|, |\muskip255|, and
|\toks255| are
traditionally kept available for such purposes.  Furthermore, plain \TeX\
reserves |\dimen0| to |\dimen9|, |\skip0| to |\skip9|, |\muskip0| to
|\muskip9|, and |\box0| to |\box9| for ``scratchwork''; these registers
are never allocated by the |\new...|\null\ operations.  We have seen that
|\count0| through |\count9| are special, and |\box255| also turns out to
be special; so those registers should be avoided unless you know what you
are doing.

\ddanger Of course any register can be used for short-term purposes inside
a group (including |\count0| to |\count9| and |\box255|, and including
registers that have been allocated for other purposes), since register
changes are local to ^{groups}. However, you should be sure that \TeX\
will not output any pages before the group has ended, because output
routines might otherwise be invoked at unfortunate times.  \TeX\ is
^^{output routines, when invoked}^^{page builder, when exercised}
liable to invoke an output routine whenever it tries to move something
from the list of recent contributions to the current page, because it
might discover a page break with $c=\infty$ then. Here is a list of the
times when that can happen: (a)~At the beginning or end of a paragraph,
provided that this paragraph is being contributed to the main vertical
list. (b)~At the beginning or end of a displayed equation within such a
paragraph. (c)~After completing an |\halign| in vertical mode. (d)~After
contributing a box or penalty or insertion to the main vertical list.
(e)~After an |\output| routine has ended.

\ddanger Now that we are armed with the knowledge of \TeX's flexible
registers, we can plunge into the details of insertions. There are 255
classes of insertions, |\insert0| to |\insert254|, and they are tied
to other registers of the same number. For example, |\insert100| is
connected with |\count100|, |\dimen100|, |\skip100|, and |\box100|.
Therefore plain \TeX\ provides an allocation function for insertions
as it does for registers; Appendix~B includes the command
\begintt
\newinsert\footins
\endtt
^^|\newinsert|
which defines |\footins| as the number for footnote insertions. Other
commands that deal with footnotes refer to |\count\footins|, |\dimen\footins|,
and so on. The macros for floating topinserts are similarly prefaced by
`|\newinsert\topins|', which defines |\topins| as the number of their
class. Each class of insertions is independent, but \TeX\ preserves the
order of insertions within a class. It turns out that |\footins| is class~254,
and |\topins| is class~253, but the macros do not use such numbers directly.

\def\n{\thinspace$n$}
\ddanger For our purposes let's consider a particular class of insertions
called class $n$; we will then be dealing with \TeX's primitive command
^^|\insert|
\begindisplay
|\insert|\n|{|\<vertical mode material>|}|
\enddisplay
which puts an insertion
item into a horizontal or vertical list. For this class of insertions
\begindisplay
|\box|\n\ is where the material appears when a page is output;\cr
|\count|\n\ is the magnification factor for page breaking;\cr
|\dimen|\n\ is the maximum insertion size per page;\cr
|\skip|\n\ is the extra space to allocate on a page.\cr
\enddisplay
For example, material inserted with |\insert100| will
eventually appear in |\box100|.

\ddanger Let the natural height plus depth of\/ |\insert|\n\ be $x$; then
|\count|\n\ is 1000 times the factor by which $x$~affects the page
goal. For example, plain \TeX\ sets |\count\footins=1000|, since
there is a one-to-one relationship:
A 10-point footnote effectively makes a page $10\pt$ shorter. But if we
have an application where footnotes appear in double columns, a count
value of 500 would be appropriate. One of the insertion classes in Appendix~E
makes marginal notes for proofreading purposes; in that case the
count value is zero. No actual magnification is done; |\count|\n\
is simply a number used for bookkeeping, when estimating the
costs of various page breaks.

\ddanger The first footnote on a page requires extra space, since we want
to separate the footnotes from the text, and since we want to output a
horizontal rule. Plain \TeX\ sets
`|\skip\footins=|^|\bigskipamount|'; this means that a bigskip of
extra space is assumed to be added by the output routine to any page that
contains at least one insertion of class |\footins|.

\ddanger Sometimes it is desirable to put a maximum limitation on the size
of insertions; for example, people usually don't want an entire page to
consist of footnotes. Plain \TeX\ sets |\dimen\footins=8in|; this means
that |\box\footins| is not supposed to accumulate more than 8~inches of
footnotes for any one page.

\ddanger You might want to review the page-breaking algorithm explained
at the beginning of this chapter, before reading further.
On the other hand, maybe you don't really want to read the rest of
this chapter at all, ever.

\ddanger Here now is the algorithm that \TeX\ performs when an |\insert|\n\
is moved from the ``recent contributions'' to the ``current page.'' \
(Remember that such a move does not mean that the insertion will actually take
place; the current page will be backed up later, to the breakpoint of
least cost, and only the insertions preceding that breakpoint will
actually be performed.) \ Let $g$ and $t$ be the current ^|\pagegoal| and
^|\pagetotal|; let $q$ be the ^|\insertpenalties| accumulated for the
current page; and let $d$ and $z$ be the current ^|\pagedepth| and
^|\pageshrink|. \ (The value of~$d$ is at most ^|\maxdepth|; this value
has not yet been incorporated into $t$.) \ Finally, let $x$ be the
natural height plus depth of the |\insert|\n\ that we are moving to the
current page; and let $f$ be the corresponding magnification factor,
i.e., |\count|\n\ divided by 1000.

\ninepoint
\textindent{\bf Step 1.} If there is no previous |\insert|\n\ on the
current page, decrease $g$ by $hf+w$, where $h$ is the current height plus
depth of\/ |\box|\n, and where $w$ is the natural space component of
|\skip|\n; also include the stretch and shrink components of
|\skip|\n\ in the totals for the current page (in particular, this
affects~$z$).

\medbreak
\textindent{\bf Step 2.} If a previous |\insert|\n\ on the current page
has been split, add the parameter called
^|\floatingpenalty| to~$q$, and omit Steps 3 and~4.

\medbreak
\textindent{\bf Step 3.} Test if the current insertion will fit on the page
without splitting. This means that it won't make the natural height-plus-depth
of\/ |\box|\n\ surpass |\dimen|\n, when it is added to |\box|\n\ together with
all previous |\insert|\n\ amounts on the current page; furthermore, it
means that either $xf\le0$ or $t+d+xf-z\le g$. If both tests are passed,
subtract $xf$ from~$g$ and omit Step~4.

\medbreak
\textindent{\bf Step 4.} (The current insertion will be split, at least
tentatively; but the split will not actually take place if the least-cost
page turns out to have occurred earlier than the present insertion.) \
First compute the largest amount~$v$ such that a height plus depth of $v$
will not make the total insertions into |\box|\n\ bigger than |\dimen|\n,
and such that $t+d+vf\le g$. \ (Notice that $z$ is omitted from the latter
formula, but the available shrinkability was considered in Step~3 when we
tried to avoid splitting.) \ Then find the least-cost way to split the
beginning of the vertical list of the insertion so as to obtain a box of
height~$v$.  \ (Use an algorithm just like page-breaking, but without the
complexity of insertion; an additional `|\penalty-10000|' item is assumed
to be present at the end of the vertical list, to ensure that a legal
breakpoint exists.) \ Let $u$ be the natural height plus depth of that
least-cost box, and let $r$ be the penalty associated with the optimum
breakpoint.  Decrease $g$ by~$uf$, and increase $q$ by~$r$. \ (If
|\tracingpages||=1|, the log file should now get a cryptic message that says
`|% split|\n\ |to| $v$|,|$u$ |p=|$r$'. For~example,^^{split insertion penalty}
\begintt
% split254 to 180.2,175.3 p=100
\endtt
^^{percent split}
means that \TeX\ has tried to split an |\insert254| to height $180.2\pt$;
the natural height-plus-depth of the best such split is $175.3\pt$,
and the penalty for breaking there is~100.)

\ddanger This algorithm is admittedly complicated, but no simpler mechanism
seems to do nearly as much. Notice that penalties of $-10000$ inside
insertions will make certain splits very attractive in Step~4, so the
user can provide hints about where to break, in difficult situations.
The algorithm provides a variety of different behaviors:
Floating insertions can be accommodated
as a special case of split insertions, by making each floating topinsert
start with a small penalty, and by having
zero as the associated |\floatingpenalty|; non-floating insertions
like footnotes are accommodated by associating larger penalties with
split insertions (see Appendix~B).

\ddanger The splitting operation mentioned in Step 4 is also available as
a primitive: `^|\vsplit|\<number> |to|\<dimen>' produces a vbox obtained
by splitting off a speci\-fied amount of material from a box register.
For example,
\begintt
\setbox200=\vsplit100 to 50pt
\endtt
sets |\box200| to a vbox whose height is $50\pt$; it goes through the
vertical list inside |\box100| (which should be a vbox) and finds the
least-cost break assuming a goal height of~$50\pt$, considering badnesses and
penalties just as in the case of page-breaking (but with $q=0$). The
algorithm uses ^|\splitmaxdepth| instead of\/ ^|\maxdepth| to govern the
maximum depth of boxes. Then it prunes the top of\/ |\box100| by removing
everything up to and including any ^{discardable} items that immediately
follow the optimum breakpoint; and it uses ^|\splittopskip| to insert new
glue before the first box inside |\box100|, just as
^|\topskip| glue appears at the top of a page. However, if the optimum
breakpoint occurs at the end of the vertical list inside |\box100|---a
`|\penalty-10000|' item is assumed to be present there---or if all items
after the optimum breakpoint are discarded, |\box100| will
be void after the |\vsplit|. And if\/ |\box100| was void before the
|\vsplit|, both |\box100| and |\box200| will be void afterwards.

\ddanger \looseness=-1
You'd better not change |\box|\n, |\count|\n, |\dimen|\n,
or |\skip|\n\ while \TeX\ is contributing insertions to the current page,
since \TeX's algorithm assumes that those quantities are static.
But you can change ^|\floatingpenalty|, |\splittopskip|, and |\splitmaxdepth|;
\TeX\ will use the values that were current just inside the closing right
brace of `|\insert|\n|{...}|' when it splits and floats insertions.
For example, Appendix~B uses |\floatingpenalty=20000| in footnote
insertions, to discourage footnotes that split before others
can start, \vadjust{\break}but |\floatingpenalty=0| in floating topinserts.
Appendix~B also uses special values of\/ |\splittopskip| and |\splitmaxdepth|,
together with ^{struts}, so that split footnotes will be typeset with
the same spacing as unsplit ones.

\ddanger The |\footnote| macro puts an |\insert| into the horizontal list
of a paragraph. After the paragraph has been broken into lines, this
insertion will move out into the vertical list just after the line that
contained it (see Chapter~14).  Since there is no legal breakpoint between
that box (i.e., that line) and the insertion, \TeX\ will put the insertion
onto the page that contains the line that contains the insertion.

\ddangerexercise Study the page-breaking algorithm carefully. Is it
possible that a footnote might not appear on the same page as its reference?
\answer Yes, in severe circumstances.  (1)~Previous footnotes might
 have left no room for any more footnotes on the page.
(2)~If |\vadjust{\eject}| occurs on the same line as a footnote, before that
footnote, the reference will be forcibly detached. (3)~Other |\vadjust|
commands on that line could also interpose breakpoints before the insertion.

\ddanger \looseness=-1
When the best page break is finally chosen, \TeX\ removes everything after
the chosen breakpoint from the bottom of the ``current page,'' and puts it
all back at the top of the ``recent contributions.'' The
chosen breakpoint itself is placed at the very top of the recent contributions.
If it is a penalty item, the value of the penalty is recorded in
^|\outputpenalty| and the penalty in the contribution list is changed
to $10000$; otherwise |\outputpenalty| is set to 10000.
The insertions that remain on the current page are of three kinds: For
each class~$n$ there are unsplit insertions, followed possibly by
a single split insertion, followed possibly by others. If
^|\holdinginserts|$\null>0$, all insertions remain in place (so that
they might be contributed again); otherwise they are all removed from
the current page list as follows: The unsplit insertions
are appended to |\box|\n, with no ^{interline glue} between them. \
(^{Struts} should be used, as in the |\vfootnote| macro of
Appendix~B\null.) \
If a split insertion is present, it is effectively |\vsplit| to the size
that was computed previously in Step~4; the top part is treated as an
unsplit insertion, and the remainder (if any) is converted to an insertion
as if it had not been split. This remainder, followed by any other floating
insertions of the same class, is held
over in a separate place. \ (They will show up on the ``current page'' if
^|\showlists| is used while an ^|\output| routine is active; the total
number of such insertions appears in ^|\insertpenalties| during an
|\output| routine.) %\
Finally, the remaining items before the best break on the current page are put
together in a |\vbox| of height~$g$, where $g$ was the |\pagegoal| at the
time of the break, using the saved value of\/ ^|\maxdepth|; this box becomes
|\box255|. Now the user's ^|\output| ^^|\box255|
routine enters \TeX's scanner (see Chapter~23); its duty is to assemble the
final pages based on the contents of\/ |\box255| and any insertion boxes
that it knows about. The output routine will probably unbox those boxes,
so that their glue can be reset; the glue in insertion boxes usually
cooperates nicely with the glue on the rest of the page, when it is
given a chance. After the |\output| routine is finished, ^{held-over
insertion} items are placed first on the list of recent contributions, followed
by the vertical list constructed by |\output|, followed by the recent
contributions beginning with the page break. \ (Deep breath.) \ You got that?

\endchapter

Since it is impossible to foresee how [footnotes] will happen to come out
in the make-up, it is impracticable to number them from 1 up on each page.
The best way is to number them consecutively throughout an article
or by chapters in a book.
\author UNIVERSITY OF ^{CHICAGO} PRESS, {\sl Manual of Style\/} (1910) % p102

\bigskip

Don't use footnotes in your books, Don.
\author JILL ^{KNUTH} (1962)

\eject
\beginchapter Chapter 16. Typing\\Math Formulas

\TeX\ is designed to handle complex ^{mathematical expressions} in such a way
that most of them are easy to input. The basic idea is that a complicated
formula is composed of less complicated ^{formulas} put together in a
simple way; the less complicated formulas are, in turn, made up of simple
combinations of formulas that are even less complicated; and so on. Stating
this another way, if you know how to type simple formulas and how to combine
formulas into larger ones, you will be able to handle virtually any formula
at all. So let's start with simple ones and work our way up.

The simplest formula is a single letter, like `$x$', or a single number,
like `2'. In order to put these into a \TeX\ text, you type `|$x$|' and
`|$2$|', respectively. Notice that all mathematical formulas are enclosed
in special math brackets; we are using |$| as the math bracket in this
manual, in accord with the plain \TeX\ format defined in Appendix~B\null,
because mathematics is supposedly expensive.

When you type `|$x$|' the `$x$' comes out in italics, but when you
type `|$2$|' the `$2$' comes out in roman type.  In general, all
characters on your keyboard have a special interpretation in math
formulas, according to the normal conventions of mathematics printing:
Letters now denote ^{italic} letters, while digits and punctuation
denote ^{roman} digits and punctuation; a hyphen ({\tt-}) now denotes a ^{minus
sign} ($-$), which is almost the same as an em-dash but not quite (see
Chapter~2). The first |$| that you type puts you into ``^{math mode}'' and the
second takes you out (see Chapter~13).  So if you forget one |$| or type
one |$| too many, \TeX\ will probably become thoroughly confused and you
will probably get some sort of error message. ^^{dollarsign}

Formulas that have been typeset by a printer who is unaccustomed to
mathematics usually look quite strange to a mathematician, because a
novice printer usually gets the spacing all wrong. In order to alleviate
this problem, \TeX\ does most of its own spacing in math formulas; and it
{\sl ignores\/} any ^{spaces} that you yourself put between |$|'s. For
example, if you type `|$ x$|' and `|$ 2 $|', they will mean the same thing
as `|$x$|' and `|$2$|'. You can type `\hbox{|$(x + y)/(x - y)$|}' or
`|$(x+y) / (x-y)$|', but both will result in `$(x+y)/(x-y)$', a formula in
which there is a bit of extra space surrounding the $+$ and~$-$ signs but
none around the~/~sign.  Thus, you do not have to memorize the complicated
rules of math spacing, and you are free to use blank spaces in any way you
like.  Of course, spaces are still used in the normal way to mark the end
of control sequences, as explained in Chapter~3.  In most circumstances
\TeX's spacing will be what a mathematician is accustomed to; but we will
see in Chapter~18 that there are control sequences by which you can
override \TeX's spacing rules if you want to.

One of the things mathematicians like to do is make their formulas look
like ^{Greek} to the uninitiated. In plain \TeX\ language you can type
`|$$\alpha, \beta, \gamma, \delta;$$|' and you will get the first four
Greek letters ^^|\alpha|^^|\beta|^^|\gamma|^^|\delta|
$$\alpha,\beta,\gamma,\delta;$$ furthermore there are uppercase Greek
letters like `$\Gamma$', which you can get by typing `|$\Gamma$|'.
^^|\Gamma| Don't feel intimidated if you aren't already familiar with Greek
letters; they will be easy to learn if you need them. The only difficulty
is that some symbols that look nearly the same must be carefully
distinguished. For example, the Greek letters ^|\nu|~($\nu$) and
^|\kappa|~($\kappa$) should not be confused with the italic letters $v$
and~$x$; the Greek ^|\phi|~($\phi$) is different from the slashed zero
called ^|\emptyset|~($\emptyset$).  A~lowercase epsilon ($\epsilon$) is
quite different from the symbol used to denote membership in a set
($\in$); type `|$\epsilon$|' for $\epsilon$ and `|$\in$|' for $\in$.
^^|\epsilon| ^^|\in| Some of the lowercase Greek letters have variant
forms in plain \TeX's math italic fonts:
`|$(\phi,\theta,\epsilon,\rho)$|' yields
`$(\phi,\theta,\epsilon,\rho)$' while
`|$(\varphi,\vartheta,\varepsilon,\varrho)$|' yields
`$(\varphi,\vartheta,\varepsilon,\varrho)$'.
^^|\phi|^^|\theta|^^|\rho|^^|\varphi|^^|\vartheta|^^|\varrho|^^|\varepsilon|

Besides Greek letters, there are a lot of ^{funny symbols} like `$\approx$'
(which you get by typing `|$\approx$|') ^^|\approx|^^{special symbols for math}
and `$\mapsto$' (which you ^^|\mapsto|^^{math symbols}
get by typing `|$\mapsto$|').
 A complete list of these control sequences
and the characters they correspond to appears in Appendix~F\null.
Such control sequences are allowed only in math mode, i.e., between
|$|'s, because the corresponding symbols appear in the math fonts.

\exercise What should you type to get the formula
`$\gamma+\nu\in\Gamma$'\thinspace?
\answer |$\gamma+\nu\in\Gamma$|.

\exercise Look at Appendix F to discover the control sequences for
`$\le$', `$\ge$', and~`$\ne$'. \ (These are probably the three most
commonly used math symbols that are not present on your keyboard.)
\ What does plain \TeX\ call them?
\answer ^|\le|, ^|\ge|, and ^|\ne|. \ (These are short for ``less-or-equal,''
``greater-or-equal,'' and ``not-equal.'') \ You can also use the names
^|\leq|, ^|\geq|, and ^|\neq|. \ (The fourth most common symbol is, perhaps,
`$\infty$', which stands for ``^{infinity}'' and is called `^|\infty|'.)

Now let's see how the more complex formulas get built up from simple ones.
In the first place, you can get ^{superscripts} $\rm^{(up\,high)}$ and
^{subscripts} $\rm_{(down\,low)}$
^^{indices, see subscripts}
^^{superiors, see superscripts} ^^{inferiors, see subscripts}
by using `|^|' and `|_|', as shown in the following examples:
\beginmathdemo
\it Input&\it Output\cr
\noalign{\vskip2pt}
|$x^2$|&x^2\cr
|$x_2$|&x_2\cr
|$2^x$|&2^x\cr
|$x^2y^2$|&x^2y^2\cr
|$x ^ 2y ^ 2$|&x ^ 2y ^ 2\cr
|$x_2y_2$|&x_2y_2\cr
|$_2F_3$|&_2F_3\cr
\endmathdemo
Notice that |^| and |_| apply only to the next single character.
If you want several things to be superscripted or subscripted, just enclose
them in braces:
\beginmathdemo
|$x^{2y}$|&x^{2y}\cr
|$2^{2^x}$|&2^{2^x}\cr
|$2^{2^{2^x}}$|&2^{2^{2^x}}\cr
|$y_{x_2}$|&y_{x_2}\cr
|$y_{x^2}$|&y_{x^2}\cr
\endmathdemo
The braces in these examples have been used to specify ``^{subformulas},''
i.e., simpler parts of a larger formula.  \TeX\ makes a box for each
subformula, and treats that box as if it were a single symbol.  Braces
also serve their usual purpose of grouping, as discussed in Chapter~5.

It is illegal to type `|x^y^z|' or `|x_y_z|'; \TeX\ will
complain of a ``double superscript'' or ``double subscript.'' You must type
`|x^{y^z}|' or `|x^{yz}|' or
`|x_{y_z}|' or `|x_{yz}|' in order to make your intention clear.

A superscript or subscript following a character applies to that character
only; but when following a subformula it applies to that whole subformula,
and it will be raised or lowered accordingly. For example,
\beginmathdemo
|$((x^2)^3)^4$|&((x^2)^3)^4\cr
|${({(x^2)}^3)}^4$|&{({(x^2)}^3)}^4\cr
\endmathdemo
In the first formula the `|^3|' and `|^4|' are superscripts on the ^{right
parentheses}, i.e., on the `|)|' characters that immediately precede them,
but in the second formula they are superscripts on the subformulas that
are enclosed in braces. The first alternative is preferable, because it is
much easier to type and it is just as easy to read.

\danger A subscript or superscript following nothing (as in the `|_2F_3|'
example on the preceding page, where the `|_2|' follows nothing) is taken
to mean a subscript or superscript of an empty subformula. Such notations
are (fortunately) rare in mathematics; but if you do encounter them it is
better to make your intention clear by showing the empty subformula
explicitly with braces.  In other words, the best way to get `${}_2F_3$'
in a formula is to type `|{}_2F_3|' or `|{_2}F_3|' or `|{_2F_3}|'.

\dangerexercise What difference, if any, is there between the output of
`|$x + _2F_3$|' and the output of `|$x + {}_2F_3$|'\thinspace?
\answer In the former, the `|_2|' applies to the plus sign ($x + _2F_3$);
but in the latter, it applies to an empty subformula ($x + {}_2F_3$).

\dangerexercise Describe the differences between the outputs of `|${x^y}^z$|'
and `|$x^{y^z}$|'.
\answer The results are `${x^y}^z$' and `$x^{y^z}$'; the $z$ in the first
alternative is the same size as the $y$, but in the second it is smaller.
Furthermore, the $y$ and $z$ in the first case aren't quite at the same height.
\ (Good typists never even think of the first construction, because
mathematicians never want it.)

You can have simultaneous subscripts and superscripts, and you can specify them
in any order:
\beginmathdemo
|$x^2_3$|&x^2_3\cr
|$x_3^2$|&x_3^2\cr
|$x^{31415}_{92}+\pi$|&x^{31415}_{92}+\pi\cr
\noalign{\smallskip}
|$x_{y^a_b}^{z_c^d}$|&x_{y^a_b}^{z_c^d}\cr
\endmathdemo
Notice that simultaneous su$\rm_b^{per\kern-1pt}$scripts are positioned
over each other.  However, a subscript will be ``tucked in'' slightly when it
follows certain letters; for example, `|$P_2^2$|' produces `$P_2^2$'.
If for some reason you want the left edges of both subscript and superscript
to be aligned, you can fool \TeX\ by inserting a null subformula:
`|$P{}_2^2$|' produces `$P{}_2^2$'.

The control sequence ^|\prime| stands for the symbol `$\prime$', which
is used mostly in superscripts. In fact, `$\prime$' is so big as it stands
that you would never want to use it except in a subscript or superscript,
where it occurs in a smaller size. Here are some typical examples:
\beginmathdemo
\it Input&\it Output\cr
\noalign{\vskip2pt}
|$y_1^\prime$|&y_1^\prime\cr
|$y_2^{\prime\prime}$|&y_2^{\prime\prime}\cr
|$y_3^{\prime\prime\prime}$|&y_3^{\prime\prime\prime}\cr
\endmathdemo
Since single and double primes occur rather frequently, plain \TeX\
provides a convenient abbreviation: You can simply type |'| instead
of |^\prime|, and |''| instead of |^{\prime\prime}|, and so on.
\beginmathdemo
|$f'[g(x)]g'(x)$|&f'[g(x)]g'(x)\cr
|$y_1'+y_2''$|&y_1'+y_2''\cr
|$y'_1+y''_2$|&y'_1+y''_2\cr
|$y'''_3+g'^2$|&y'''_3+g'{}^2\cr
\endmathdemo

\dangerexercise Why do you think \TeX\ treats |\prime| as a large symbol
that appears only in superscripts, instead of making it a smaller
symbol that has already been shifted up into the superscript position?
\answer The second alternative doesn't work properly when there's a
subscript at the same time as a prime. Furthermore, some mathematicians
use |\prime| also in the subscript position; they write, for example,
$F'(w,z)=\partial F(w,z)/\partial z$ and $F_\prime(w,z)=\partial F(w,z)/
\partial w$.

\dangerexercise Mathematicians sometimes use ``^{tensor notation}''
in which subscripts and superscripts are staggered, as in `$R_i{}^{jk}{}_l$'.
Explain how to achieve such an effect.
\answer |$R_i{}^{jk}{}_l$|.

Another way to get complex formulas from simple ones is to use the control
sequences ^|\sqrt|, ^|\underline|, or ^|\overline|.
^^{surds, see sqrt} ^^{vinculum, see overline}
Like |^| and |_|, these operations apply to the character or subformula
that follows them:
\beginmathdemo
|$\sqrt2$|&\sqrt2\cr
|$\sqrt{x+2}$|&\sqrt{x+2}\cr
|$\underline4$|&\underline4\cr
|$\overline{x+y}$|&\overline{x+y}\cr
|$\overline x+\overline y$|&\overline x+\overline y\cr
|$x^{\underline n}$|&x^{\underline n}\cr
|$x^{\overline{m+n}}$|&x^{\overline{m+n}}\cr
|$\sqrt{x^3+\sqrt\alpha}$|&\sqrt{x^3+\sqrt\alpha}\cr
\endmathdemo
You can also get cube roots `$\root3\of{\phantom{h}}$' and similar things
by using ^|\root|:
\beginmathdemo
|$\root 3 \of 2$|&\root 3 \of 2\cr
|$\root n \of {x^n+y^n}$|&\root n \of {x^n+y^n}\cr
|$\root n+1 \of a$|&\root n+1 \of a\cr
\endmathdemo

\danger The |\sqrt| and |\underline| and |\overline| operations are able to
place lines above or below subformulas of any size or shape; the bar lines
change their size and position, so that they are long enough to cover the
subformula, and high enough or low enough not to bump into it. For example,
consider `|\overline|~|l|' ($\,\overline l\,$) versus `|\overline|~|m|'
($\,\overline m\,$):  The first has a shorter bar line, and this line has
been raised higher than the bar in the second. Similarly, the bar in
`|\underline|~|y|' ($\,\underline y\,$) is lower than the bar in
`|\underline|~|x|' ($\,\underline x\,$); and square root signs appear in a
variety of positions based on the height and depth of what is being
|\sqrt|'d: $\sqrt a + \sqrt d + \sqrt y$. \TeX\ knows the height, depth, and
width of every letter and every subformula, because it considers them to be
boxes, as explained in Chapter~11. If you have a formula in which there is
only one |\sqrt|, or only one |\overline| or |\underline|, the normal
positioning rules work fine; but sometimes you want to have uniformity
between different members of a complex formula. For example, you might want to
typeset `$\mskip-1mu\sqrt{\mathstrut a}+
 \sqrt{\mathstrut d}+\sqrt{\mathstrut y}\mskip+2mu$',
putting all square roots in the same vertical position. There's an easy way
to do this, using the control sequence ^|\mathstrut| as follows:
\begintt
$\sqrt{\mathstrut a}+\sqrt{\mathstrut d}+\sqrt{\mathstrut y}$.
\endtt
A |\mathstrut| is an invisible box whose width is zero; its height and depth
are the height and depth of a parenthesis `('. Therefore subformulas
that contain |\mathstrut| will always have the same height and depth,
unless they involve more complicated constructions like subscripts and
superscripts. Chapter~18 discusses more powerful operations called ^|\smash|
and ^|\phantom| by which you can obtain complete control over the positioning
of roots and similar signs.

\exercise Test your understanding of what you have read so far in this chapter
by explaining what should be typed to get the following formulas. \ (Be
sure to check your answer with Appendix~A to confirm that you're right.)
$$\hbox to\hsize{\indent
  $\displaystyle
  10^{10}\hfil 2^{n+1}\hfil (n+1)^2\hfil \sqrt{1-x^2}\hfil
  \overline{w+\overline z}\hfil p_1^{e_1}\hfil a_{b_{c_{d_e}}}\hfil
  \root3\of{h''_n(\alpha x)}\hfil$}$$
\answer |10^{10}|; \stretch|2^{n+1}|; \stretch|(n+1)^2|; \stretch|\sqrt{1-x^2}|;
\stretch|\overline{w+\overline z}|; \stretch|p_1^{e_1}|; \stretch
|a_{b_{c_{d_e}}}|; \stretch|\root3\of{h''_n(\alpha x)}|.
\ (Of course, you should enclose these formulas in dollar signs so that
\TeX\ will process them in math mode.  Superscripts and subscripts can be
given in either order; for example, |h''_n| and |h_n''| both work the
same.  You should not leave out any of the braces shown here; for example,
`|$10^10$|' would yield `$10^10$'. But it doesn't hurt to insert
additional braces around letters or numbers, as in `|({n}+{1})^{2}|'. The
indicated blank spaces are necessary unless you use extra braces;
otherwise \TeX\ will complain about undefined control sequences
|\overlinez| and |\alphax|.)

\exercise What mistake did B. C. ^{Dull} discover after he typed the
following?
\begintt
If$ x = y$, then $x$ is equal to $y.$
\endtt
\answer He got `If$ x = y\ldots$' because he forgot to leave a space
after `|If|'; ^{spaces} disappear between dollar signs. He should
also have ended the sentence with `|$y$.|'; punctuation that belongs
to a sentence should not be included in a formula, as we will see
in Chapter~18. \ (But you aren't expected to know that yet.)

\exercise Explain how to type the following sentence:
\begindisplay
Deleting an element from an $n$-tuple leaves an $(n-1)$-tuple.
\enddisplay
\answer |Deleting an element from an $n$-tuple leaves an $(n-1)$-tuple.|

\exercise List all the italic letters that descend below the baseline.
\ (These are the letters for which |\underline| will lower its bar line.)
\answer $Q,f,g,j,p,q,y$. \ (The analogous ^{Greek} letters are
^^{italic letters with descenders} ^^{descenders}
$\beta,\gamma,\zeta,\eta,\mu,\xi,\rho,\phi,\varphi,\chi,\psi$.)

We have discussed the fact that the characters you type have special meanings
in math mode, but the examples so far are incomplete; they don't reveal all
the power that is at your fingertips just after you press the `|$|' key.
It's time now to go back to basics: Let us make a systematic survey of
what each character does, when it is used in a formula.

The 52 ^{letters} (|A| to |Z| and |a| to |z|) denote italic symbols
($A$~to~$Z$ and $a$~to~$z$), which a mathematician would call ``^{variables}.''
\TeX\ just calls them ``^{ordinary symbols},'' because they make up the
bulk of math formulas. There are two variants of lowercase L in plain \TeX,
namely `$l$' (which you get by simply typing `|l|') and `$\ell$'
(which you get by typing `^|\ell|'). Although mathematicians commonly
write something that looks like `$\ell$' in their manuscripts, they
do so only to distinguish it from the numeral~`1'.  This
distinguishability problem is not present in printed mathematics, since an
italic `$l$' is quite different from a~`1'; therefore it is traditional to
use `$l$' unless `$\ell$' has been specifically requested.

Plain \TeX\ also treats the 18 characters
\begintt
0 1 2 3 4 5 6 7 8 9 ! ? . || / ` @ "
\endtt
^^{digits}^^{numerals}
as ordinary symbols; i.e., it doesn't insert any extra space when these
symbols occur next to each other or next to letters. Unlike the letters,
these 18 characters remain in roman type when they appear in formulas.
There's nothing special for you to remember about them, except that the
^{vertical line} `\|' has special uses that we shall discuss later.
Furthermore, you should be careful to distinguish between `oh' and `zero':
The italic letter~$O$ is almost never used in formulas unless it appears
just before a left parenthesis, as in `$O(n)$'; ^^{big-$O$ notation}
and the numeral~$0$ is almost never used just before a left parenthesis
unless it is preceded by another digit, as in `$10(n-1)$'. Watch for
left parentheses and you'll be $0K$. \ (Lowercase o's also tend to
appear only before left parentheses; type `|x_0|' instead of `|x_o|',
since the formula `$x_0$' is generally more correct than `$x_o$'.)

The three characters |+|, |-|, and |*|
are called ``^{binary operations},'' because they operate on two parts of
a formula. For example, |+|~is a ^{plus sign}, which is used for the sum
of two numbers; |-|~is a ^{minus sign}. The ^{asterisk}~(|*|) is rarer
^^{star, see asterisk}
in mathematics, but it also behaves as a binary operation. Here are
some examples of how \TeX\ typesets binary operations when they appear
next to ordinary symbols:
\beginmathdemo
\it Input&\it Output\cr
\noalign{\vskip2pt}
|$x+y-z$|&x+y-z\cr
|$x+y*z$|&x+y*z\cr
|$x*y/z$|&x*y/z\cr
\endmathdemo
Notice that |-| and |*| produce quite different math symbols from what you
get in normal text: The ^{hyphen}~(-) becomes a minus sign~($-$), and
the raised asterisk~(*) drops down to a lower level~($*$).

\danger \TeX\ does not treat |/| as a binary operation, even though a
^{slash} stands for division (which qualifies as a binary operation on
mathematical grounds). The reason is that printers traditionally put extra
space around the symbols $+$, $-$, and~$*$, but not around~$/$. If \TeX\ were to
typeset |/| as a binary operation, the formula `|$1/2$|' would come out
`$1\mathbin/2$', which is wrong; so \TeX\ considers |/| to be an ordinary
symbol.

\danger Appendix F lists many more binary operations, for which you type
control sequences instead of single characters. Here are some examples:
\beginmathdemo
|$x\times y\cdot z$|&x\times y\cdot z\cr
|$x\circ y\bullet z$|&x\circ y\bullet z\cr
|$x\cup y\cap z$|&x\cup y\cap z\cr
|$x\sqcup y\sqcap z$|&x\sqcup y\sqcap z\cr
|$x\vee y\wedge z$|&x\vee y\wedge z\cr
|$x\pm y\mp z$|&x\pm y\mp z\cr
\endmathdemo
It is important to distinguish $\times$ (^|\times|) from $X$ (|X|)
and from $x$ (|x|); to distinguish $\cup$ (^|\cup|) from $U$ (|U|)
and from $u$ (|u|); to distinguish $\vee$ (^|\vee|) from $V$ (|V|)
and from $v$ (|v|); to distinguish $\circ$ (^|\circ|) from $O$ (|O|)
and from $o$ (|o|). ^^|\cdot|^^|\bullet|^^|\cap|^^|\sqcup|^^|\sqcap|^^|\wedge|
^^{cross, see dagger, times}
^^|\pm|^^|\mp| The symbols `$\lor$' and `$\land$' can also be called
^|\lor| and ^|\land|, since they frequently stand for binary operations
that are called ``^{logical~or}'' and ``^{logical~and}.''

\danger Incidentally, binary operations are treated as ordinary symbols
if they don't occur between two quantities that they can operate on.
For example, no extra space is inserted next to the $+$, $-$, and~$*$
in cases like the following:
\beginmathdemo
|$x=+1$|&x=+1\cr
|$3.142-$|&3.142-\cr
|$(D*)$|&(D*)\cr
\endmathdemo
Consider also the following examples, which show that binary
operations can be used as ordinary symbols in superscripts and subscripts:
\beginmathdemo
|$K_n^+,K_n^-$|&K_n^+,K_n^-\cr
|$z^*_{ij}$|&z^*_{ij}\cr
|$g^\circ \mapsto g^\bullet$|&g^\circ \mapsto g^\bullet\cr
|$f^*(x) \cap f_*(y)$|&f^*(x) \cap f_*(y)\cr
\endmathdemo

\dangerexercise How would you obtain the formulas `$z^{*2}$'
and `$h_*'(z)$'\thinspace?
\answer |$z^{*2}$| and |$h_*'(z)$|.

Plain \TeX\ treats the four characters |=|, |<|, |>|, and |:|\ as
``^{relations}'' because they express a relationship between two
quantities. For example, `${x<y}$' means that $x$~is less than~$y$.
Such relationships have a rather different meaning from binary
operations like $+$, and the symbols are typeset somewhat differently:
\beginmathdemo
|$x=y>z$|&x=y>z\cr
|$x:=y$|&x:=y\cr
|$x\le y\ne z$|&x\le y\ne z\cr
|$x\sim y\simeq z$|&x\sim y\simeq z\cr
|$x\equiv y\not\equiv z$|&x\equiv y\not\equiv z\cr
|$x\subset y\subseteq z$|&x\subset y\subseteq z\cr
\endmathdemo
^^|\le|^^|\ne|^^|\simeq|^^{colon}^^{equals}^^{lessthan}^^{greaterthan}
^^{colonequals}^^|\equiv|^^|\not|^^|\subset|^^|\subseteq|^^|\sim|
^^{hooks, see subset, supset} ^^{wiggle, see sim}
(The last several examples show some of the many other
relational symbols that plain \TeX\ makes available via control sequences;
see Appendix~F.)

The two characters `|,|' (^{comma}) and `|;|' (^{semicolon}) are treated
as ^{punctuation marks in formulas}; this means that \TeX\ puts a little
extra space after them, but not before them.
\beginmathdemo
|$f(x,y;z)$|&f(x,y;z)\cr
\endmathdemo
It isn't customary to put extra space after a `|.|'\ (^{period}) in
math formulas, so \TeX\ treats a period as an ordinary symbol.
If you want the `|:|'\ character to be treated as a punctuation mark
instead of as a relation, just call it ^|\colon|:
\beginmathdemo
|$f:A\to B$|&f:A\to B\cr
|$f\colon A\to B$|&f\colon A\to B\cr
\endmathdemo
^^|\to|
If you want to use a comma as an ordinary symbol (e.g., when it
appears in a large number), just put it in braces; \TeX\ treats
anything in braces as an ordinary symbol. For instance,
\beginmathdemo
|$12,345x$|&12,345x\qquad\rm(wrong)\cr
|$12{,}345x$|&12{,}345x\qquad\,\rm(right)\cr
\endmathdemo

\dangerexercise What's an easy way to get a raised dot in a decimal
constant (e.g., `$3{\cdot}1416$')?
\answer |$3{\cdot}1416$|. \ (One of the earlier examples in this
chapter showed that ^|\cdot| is a binary operation; putting it in braces
makes it act like an ordinary symbol.)\par
If you have lots of constants like this, for example in a table, there's a way
to make ordinary periods act like |\cdot| symbols: Just define
^|\mathcode||`.| to be |"0201|, assuming that the fonts of plain \TeX\ are
being used.  However, this could be dangerous, since ordinary
periods are used frequently in displayed equations; the |\mathcode| change
should be confined to places where every period is to be a |\cdot|.

So far we have considered letters, other ordinary symbols, binary operations,
relations, and punctuation marks; hence we have covered almost every key on
the typewriter. There are just a few more: The characters `|(|' ^^{lparen}
and `|[|' ^^{lbracket} are called ``^{openings},'' while `|)|' ^^{rparen} and
^^{fences, see opening, closing, delimiters}
`|]|' ^^{rbracket} are called ``^{closings}''; these act pretty much like
ordinary symbols, but they help \TeX\ to decide when a binary operation is
not really being used in a binary way.  Then there is the character~|'|,
which we know is used as an abbreviation for |\prime| superscripts.
Finally, we know that plain \TeX\ reserves the other ten characters:
\begintt
\ $ % # & ~ { } _ ^
\endtt
These are not usable for symbols in math mode unless their ^|\catcode|
values are changed (see Chapter~7). Although |{| and |}| specify
grouping, the control sequences `|\{|' and `|\}|' ^^|\lbrace|^^|\rbrace| can
be used to get `$\{$' as an opening and `$\}$' as a closing.

\ddanger All of these math mode interpretations are easily changeable, since
each character has a ^|\mathcode|, as explained in Chapter~17; none of
the conventions are permanently built into \TeX\null. However, most of them are
so standard that it is usually unwise to make many changes, except perhaps
in the interpretations of |`|, |"|, and |@|. ^^{at sign} ^^{leftquote}
^^{doublequote}

The special characters |^| and |_| that designate superscripts
^^{circumflex}^^{underbar}
and subscripts should not be used except in formulas. Similarly,
the names of math symbols like |\alpha| and |\approx|, and the
control sequences for math operations like |\overline|, must not
invade ordinary text. \TeX\ uses these facts to detect ^{missing dollar
signs} in your input, before such mistakes cause too much trouble. For
example, suppose you were to type
\begintt
The smallest $n such that $2^n>1000$ is~10.
\endtt
\TeX\ doesn't know that you forgot a `|$|' after the first `|n|', because
it doesn't understand English; so it finds a ``formula'' between the
first two |$| signs:
\begindisplay
The smallest $n such that $
\enddisplay
after which it thinks that `|2|' is part of the text. But then the |^|
reveals an inconsistency; \TeX\ will automatically insert a~|$| before
the~|^|, and you will get an error message. In this way the computer has
gotten back into synch, and the rest of the document can be typeset as if
nothing had happened.

\danger Conversely, a blank line or ^|\par| is not permitted in math mode.
This gives \TeX\ another way to recover from a missing~|$|; such
errors will be confined to the paragraph in which they occur.

\danger If for some reason you cannot use |^| and |_| for superscripts
and subscripts, because you have an unusual keyboard or because you need
|^| for French accents or something, plain \TeX\ lets you type ^|\sp|
and ^|\sb| instead. For example, `|$x\sp2$|' is another way to get `$x\sp2$'.
On the other hand, some people are lucky enough to have keyboards that
contain additional symbols besides those of standard ASCII.
^^{character set}
When such symbols are available, \TeX\ can be set up to make math
typing a bit more pleasant. For example, at the author's installation there
are keys labeled \up\ and~\dn\ that produce visible symbols
(these make superscripts and subscripts look much nicer
on the screen); there are keys for the relations {\tentex\char'34},
{\tentex\char'35}, and {\tentex\char'32} (these save time); and there are
about two dozen more keys that occasionally come in handy. \ (See Appendix~C.)
^^{uparrow}^^{downarrow}^^{leq}^^{geq}^^{neq}

\danger Mathematicians are fond of using ^{accents} over letters, because
this is often an effective way to indicate relationships between
mathematical objects, and because it greatly extends the number of available
symbols without increasing the number of necessary fonts.
Chapter~9 discusses the use of accents in ordinary text, but mathematical
accents are somewhat different, because spacing is not the same; \TeX\ uses
special conventions for accents in formulas, so that the two sorts of
accents will not be confused with each other. The following math accents
are provided by plain~\TeX:
\beginmathdemo
|$\hat a$|&\hat a\cr
|$\check a$|&\check a\cr
|$\tilde a$|&\tilde a\cr
|$\acute a$|&\acute a\cr
|$\grave a$|&\grave a\cr
|$\dot a$|&\dot a\cr
|$\ddot a$|&\ddot a\cr
|$\breve a$|&\breve a\cr
|$\bar a$|&\bar a\cr
|$\vec a$|&\vec a\cr
\endmathdemo
^^|\hat|^^|\check|^^|\tilde|^^|\acute|^^|\grave|
^^|\dot|^^|\ddot|^^|\breve|^^|\bar|^^|\vec|
The first nine of these are called |\^|, |\v|, |\~|, |\'|, |\`|, |\.|,
|\"|, |\u|, and |\=|, respectively, when they appear in text; |\vec| is an
accent that appears only in formulas. \TeX\ will complain if you try to
use |\^| or |\v|, etc., in formulas, or if you try to use |\hat| or
|\check|, etc., in ordinary text.

\danger It's usually a good idea to define special control sequences for
accented letters that you need frequently. For example, you can put
^^|\def|
\begintt
\def\Ahat{{\hat A}}
\def\chat{{\hat c}}
\def\scheck{{\check s}}
\def\xtilde{{\tilde x}}
\def\zbar{{\bar z}}
\endtt
at the beginning of a manuscript that uses the symbols $\hat A$, $\hat c$,
$\check s$, $\tilde x$, and $\bar z$ more than, say, five times.  This
saves you a lot of keystrokes, and it makes the manuscript easier to
read.  Chapter~20 explains how to define control sequences.

\danger When the letters $i$ and $j$ are accented in math formulas, ^{dotless}
symbols $\imath$ and $\jmath$ should be used under the accents. These
symbols are called ^|\imath| and ^|\jmath| in plain \TeX. Thus, for example,
a paper that uses `$\hat\imath$' and `$\hat\jmath$' ought to begin with
the following definitions:
\begintt
\def\ihat{{\hat\imath}}
\def\jhat{{\hat\jmath}}
\endtt

\def\Ahat{{\hat A}}
\danger You can put ^{accents on top of accents}, making symbols like
$\skew6\hat\Ahat$ that might cause a mathematician to
squeal with ecstasy. However, it takes a bit of finesse to get the
upper accent into a position that looks right, because the designer of a
font for mathematics usually tells \TeX\ to position math accents in
special ways for special letters. Plain \TeX\ provides a control sequence
called ^|\skew| that makes it fairly easy to shift superaccents into their
proper place. For example, `|\skew6\hat\Ahat|' was used to produce the
symbol above. The number `|6|' in this example was chosen by trial and
error; `|5|'~seems to put the upper accent a bit too far left, while
`|7|'~makes it a bit too far right, at least in the author's opinion.
The idea is to fiddle with the amount of skew until you find what
pleases you best.

\danger It's possible, in fact, to put math accents on any subformula, not
just on single characters or accented characters. But there's usually not
much point in doing so, because \TeX\ just centers the accent over the
whole subformula. For example, `|$\hat{I+M}$|' yields `$\hat{I+M}$'. In
particular, a |\bar| accent always stays the same size; it's not like
^|\overline|, which grows with the formula under it. Some people prefer
the longer line from |\overline| even when it applies to only a single
letter; for example, `|$\bar z+\overline z$|' produces `$\bar z+\overline
z$', and you can take your pick when you define |\zbar|.  However, plain
\TeX\ does provide two accents that grow; they are called ^|\widehat| and
^|\widetilde|:
\beginmathdemo
|$\widehat x,\widetilde x$|&\tenmath\widehat x,\widetilde x\cr
|$\widehat{xy},\widetilde{xy}$|&\tenmath\widehat{xy},\widetilde{xy}\cr
|$\widehat{xyz},\widetilde{xyz}$|&\tenmath\widehat{xyz},\widetilde{xyz}\cr
\endmathdemo
The third example here shows the maximum size available.

\def\ghat{{\hat g}}
\exercise This has been another long chapter; but cheer up, you have learned
a lot! Prove it by explaining what to type in order to get the formulas
$e^{-x^2}$, $D\sim p^\alpha M+l$, and $\ghat\in(H^{\pi_1^{-1}})'$. \ (In
the last example, assume that a control sequence |\ghat| has already been
defined, so that |\ghat| produces the accented letter $\ghat$.)
\answer |$e^{-x^2}$|, |$D\sim p^\alpha M+l$|, and |$\ghat\in(H^{\pi_1^{-1}})'$|.
\ (If you are reading the dangerous bend sections, you know that the
recommended way to define |\ghat| is `|\def\ghat{{\hat g}}|'.)

\endchapter

Producing ^{Greek} letters is as easy as $\pi$.
You just type |... as easy as $\pi$.|
\author LESLIE ^{LAMPORT}, {\sl The ^{L\kern-.2em\raise.6ex\hbox{a}%
  \kern-.1em\TeX} Document Preparation System\/} (1983)
        % Note: the final manual has a slightly different wording on p43.
        % It's now called "LaTeX: A Document Preparation System" (1986)
        % But I decided to cite the original, partly because I have
        % no smallcaps sans-serif `A' to match the new LaTeX logo!

\bigskip

\TeX\ has no regard for the glories of the Greek tongue---\/
as far as it is concerned, Greek letters are just additional weird symbols,
and they are allowed\/ {\rm only} in math mode.
In a pinch you can get the output $\tau\epsilon\chi$ by typing %
  |$\tau\epsilon\chi$|,
but if you're actually setting Greek text, you will be using
a different version of \TeX, designed for a keyboard with Greek letters on it,
and you shouldn't even be reading this manual,
which is undoubtedly all English to you.
\author MICHAEL ^{SPIVAK}, {\sl The Joy of \TeX\/} (1982)

\eject
\beginchapter Chapter 17. More about Math

Another thing mathematicians like to do is make fractions---and they
like to build symbols up on top of each other in a variety of different ways:
\begindisplay
$\displaystyle
{1\over2}\qquad{\rm and}\qquad{n+1\over3}\qquad{\rm and}\qquad
{n+1\choose3}\qquad{\rm and}\qquad\sum_{n=1}^3 Z_n^2\,.$
\enddisplay
You can get these four formulas as displayed equations by typing
`|$$1\over2$$|' and
`|$$n+1\over3$$|' and
`|$$n+1\choose3$$|' and
`|$$\sum_{n=1}^3 Z_n^2$$|';
we shall study the simple rules for such constructions in this chapter.
^^|\sum|^^|\choose|

First let's look at ^{fractions}, which use the `^|\over|' notation. The
control sequence |\over| applies to everything in the formula unless you
use braces to enclose it in a specific subformula; in the latter
^^{stacked fractions, see over}
case, |\over| applies to everything in that subformula.
\begindisplaymathdemo
\it Input&\it Output\cr
\noalign{\vskip-3pt}
|$$x+y^2\over k+1$$|&x+y^2\over k+1\cr
\noalign{\vskip2pt}
|$${x+y^2\over k}+1$$|&{x+y^2\over k}+1\cr
\noalign{\vskip-1pt}
|$$x+{y^2\over k}+1$$|&x+{y^2\over k}+1\cr
\noalign{\vskip-1pt}
|$$x+{y^2\over k+1}$$|&x+{y^2\over k+1}\cr
\noalign{\vskip-3pt}
|$$x+y^{2\over k+1}$$|&x+y^{2\over k+1}\cr
\endmathdemo
You aren't allowed to use |\over| twice in the same subformula; instead of
typing something like
`|a \over b \over 2|', you must specify what goes over what:
\begindisplaymathdemo
\noalign{\vskip3pt}
|$${a\over b}\over 2$$|&{a\over b}\over 2\cr
|$$a\over{b\over 2}$$|&a\over{b\over 2}\cr
\endmathdemo
Unfortunately, both of these alternatives look pretty awful. Mathematicians
tend to ``overuse'' |\over| when they first begin to typeset their own work
on a system like \TeX. A good typist or copy editor will convert fractions
to a ``^{slashed form},'' whenever a built-up construction would be too
small or too crowded. For example, the last two cases should be treated
as follows:
\begindisplaymathdemo
\noalign{\vskip3pt}
|$$a/b \over 2$$|&a/b \over 2\cr
|$$a \over b/2$$|&a \over b/2\cr
\endmathdemo
Conversion to slashed form takes a little bit of mathematical knowhow, since
^{parentheses} sometimes need to be inserted in order to preserve the meaning
of the formula. Besides substituting `|/|' for~`|\over|', the two parts
of the fraction should be put in parentheses unless they are single
symbols; for example, $a\over b$~becomes simply~$a/b$, but
$a+1\over b$ becomes $(a+1)/b$, and $a+1\over b+1$ becomes
${(a+1)/(b+1)}$. Furthermore, the entire fraction should generally
be enclosed in parentheses if it appears next to something else;
for example, ${a\over b}x$ becomes $(a/b)x$. If you are a typist without
mathematical training, it's best to ask the author of the manuscript
for help, in doubtful cases; you might also tactfully suggest that
unsightly fractions be avoided altogether in future manuscripts.

\exercise What's a better way to render the formula $x+y^{2\over k+1}$?
\answer $x+y^{2/(k+1)}$\quad(|$x+y^{2/(k+1)}$|).

\exercise Convert `${a+1\over b+1}x$' to slashed form.
\answer $((a+1)/(b+1))x$\quad(|$((a+1)/(b+1))x$|).

\exercise What surprise did B. L. ^{User} get when he typed `|$$x = (y^2\over
k+1)$$|'\thinspace?
\answer He got the displayed formula$$x=(y^2\over k+1)$$ because he forgot
that an unconfined |\over| applies to everything.  \ (He should probably
have typed `|$$x=\left(y^2\over k+1\right)$$|', using ideas that will be
presented later in this chapter; this not only makes the parentheses
larger, it keeps the `$x=$' out of the fraction, because |\left| and
|\right| introduce subformulas.)

\def\cents{\hbox{\rm\rlap/c}}
\exercise How can you make `$7{1\over2}\cents$'? \ (Assume that
the control sequence |\cents| yields~`$\cents$'.)^^{money}^^{cents}
\answer `|$7{1\over2}\cents$|' or `|7$1\over2$\cents|'. \ (Incidentally,
the definition used here was |\def\cents{\hbox{\rm\rlap/c}}|.)
^^|\rlap|^^|\cents|

The examples above show that letters and other symbols sometimes get
smaller when they appear in fractions, just as they get smaller when they
are used as exponents. It's about time that we studied \TeX's method for
choosing the sizes of things. \TeX\ actually has eight different
^{styles} in which it can treat formulas, namely
$$\halign{\indent#\hfil\quad&#\hfil\cr
display style&(for formulas displayed on lines by themselves)\cr
text style&(for formulas embedded in the text)\cr
script style&(for formulas used as superscripts or subscripts)\cr
scriptscript style&(for second-order superscripts or subscripts)\cr}$$
^^{display style}^^{text style}^^{script style}^^{scriptscript style}
and four other ``^{cramped}'' styles that are almost the same except that
exponents aren't raised quite so much. For brevity we shall refer to the
eight styles as
\begindisplay
$\displaystyle D,\ D',\ T,\ T',\ S,\ S',\ \SS,\ \SS',$
\enddisplay
where $D$ is display style, $D'$ is cramped display style, $T$~is text style,
etc. \TeX\ also uses three different ^{sizes of type for mathematics};
they are called ^{text size}, ^{script size}, and ^{scriptscript size}.

The normal way to typeset a formula with \TeX\ is to enclose it in dollar
signs |$|$\,\ldots\,$|$|; this yields the formula in text style
(style~$T$). Or you can enclose it in double dollar signs |$$|$\,\ldots\,$|$$|;
this displays the formula in display style (style~$D$). The subformulas of
a formula might, of course, be in different styles. Once you know
the style, you can determine the size of type that \TeX\ will use:
$$\everycr{\noalign{\penalty10000}}
\halign{\indent#\hfil\qquad&#\hfil&\quad#\llap(like this)\hfil\cr
If a letter is in style&then it will be set in\cr
\noalign{\vskip 2pt}
$D,D',T,T'$&text size&\cr
$S,S'$&script size&\sevenrm\cr
$\SS,\SS'$&scriptscript size&\fiverm\cr}$$
There is no ``$\it SSS$'' style or ``scriptscriptscript'' size; such tiny
symbols would be even less readable than the scriptscript ones. Therefore
\TeX\ stays with scriptscript size as the minimum:
$$\halign{\indent\hbox to 1.3in{#\hfil}&\hbox to 1.2in{#\hfil}&#\hfil\cr
In a formula&the superscript&and the subscript\cr
of style&style is&style is\cr
\noalign{\vskip 2pt}
$D,T$&$S$&$S'$\cr
$D',T'$&$S'$&$S'$\cr
$S,\SS$&$\SS$&$\SS'$\cr
$S',\SS'$&$\SS'$&$\SS'$\cr}$$
For example, if |x^{a_b}| is to be typeset in style $D$, then |a_b| will
be set in style~$S$, and {\tt b}~in style~$\SS'$; the result is
`$\displaystyle x^{a_b}$'.

So far we haven't seen any difference between styles $D$ and $T$. Actually
there is a slight difference in the positioning of exponents, although
script size is used in each case: You get
$\displaystyle x^2$~in $D$~style and $x^2$~in $T$~style and \vbox to 0pt{
\vss\hbox{$\displaystyle{\atop x^2}$}\kern0pt}~in $D'$ or $T'$~style---do
you see the difference? But there is a big distinction between $D$ style and
$T$ style when it comes to fractions:
$$\halign{\indent\hbox to 1.3in{#\hfil}&\hbox to 1.2in{#\hfil}&#\hfil\cr
In a formula&the style of the&and the style of the\cr
$\alpha$|\over|$\,\beta$ of style&numerator $\alpha$ is&denominator
$\beta$ is\cr
\noalign{\vskip 2pt}
$D$&$T$&$T'$\cr
$D'$&$T'$&$T'$\cr
$T$&$S$&$S'$\cr
$T'$&$S'$&$S'$\cr
$S,\SS$&$\SS$&$\SS'$\cr
$S',\SS'$&$\SS'$&$\SS'$\cr}$$
^^{numerator}^^{denominator}
Thus if you type `|$1\over2$|' (in a text) you get $1\over2$, namely style
$S$ over style~$S'$; but if you type
`|$$1\over2$$|' you get $$1\over2$$ (a displayed formula), which is style
$T$ over style $T'$.

\danger While we're at it, we might as well finish the style rules:
^|\underline| does not change the style. ^{Math accents}, and the operations
^|\sqrt| and ^|\overline|, change uncramped styles to their cramped
counterparts; for example, $D$ changes to $D'$, but $D'$ stays as it was.

\dangerexercise State the style and size of each part of the formula
$\displaystyle \sqrt{p_2^{e'}}$, assuming that the formula itself is in
style~$D$.
\answer Style $D'$ is used for the subformula $p_2^{e'}$, hence style~$S'$
is used for the superscript~$e'$ and the subscript~2, and style~$\SS'$
is used for the supersuperscript prime. The square root sign and the $p$
appear in text size; the 2 and the~$e$ appear in script size; and the
$\prime$ is in scriptscript size.

Suppose you don't like the style that \TeX\ selects by its automatic style
rules.  Then you can specify the style you want by typing ^|\displaystyle|
or ^|\textstyle| or ^|\scriptstyle| or ^|\scriptscriptstyle|; the style
that you select will apply until the end of the formula or subformula, or
until you select another style.  For example,
`|$$n+\scriptstyle n+\scriptscriptstyle n.$$|' produces the display
$$n+\scriptstyle n+\scriptscriptstyle n.$$
This is a rather silly example, but it does show
that the plus signs get smaller too, as the style changes. \TeX\ puts no
space around + signs in script styles.

Here's a more useful example of style changes: Sometimes you need to
typeset a ``^{continued fraction}'' made up of many other fractions,
all of which are supposed to be in display style:
$$a_0+{1\over\displaystyle a_1+
          {\strut 1\over\displaystyle a_2+
            {\strut 1\over\displaystyle a_3+
              {\strut 1\over a_4}}}}$$
In order to get this effect, the idea is to type
\begintt
$$a_0+{1\over\displaystyle a_1+
        {\strut 1\over\displaystyle a_2+
          {\strut 1\over\displaystyle a_3+
            {\strut 1\over a_4}}}}$$
\endtt
(The control sequence ^|\strut| has been used to make the denominators
taller; this is a refinement that will be discussed in
Chapter~18. Our concern now is with the style commands.) \
Without the appearances of\/ |\strut| and |\displaystyle| in this formula,
the result would be completely different:
$$a_0+{1\over a_1+{1\over
      a_2+{1\over a_3+{1\over a_4}}}}$$

\danger These examples show that the numerator and denominator of a fraction
are generally centered with respect to each other. If you prefer to have
the numerator or denominator appear ^{flush left}, put `^|\hfill|' after
it; or if you prefer ^{flush right}, put `|\hfill|' at the left. For
example, if the first three appearances of `|1\over|' in the previous
example are replaced by `|1\hfill\over|', you get the display
$$a_0+{1\hfill\over\displaystyle a_1+
          {\strut1\hfill\over\displaystyle a_2+
            {\strut1\hfill\over\displaystyle a_3+
              {\strut1\over a_4}}}}$$
(a format for continued fractions that many authors prefer). This works
because |\hfill| stretches at a faster rate than the glue that is
actually used internally by \TeX\ when it centers the numerators
and denominators.

\TeX\ has another operation `^|\atop|', which is like |\over| except that
it leaves out the fraction line:
\begindisplaymathdemo
|$$x\atop y+2$$|&x\atop y+2\cr
\endmathdemo
The plain \TeX\ format in Appendix B also defines `^|\choose|', which is
like |\atop| but it encloses the result in parentheses:
\begindisplaymathdemo
|$$n\choose k$$|&n\choose k\cr
\endmathdemo
It is called |\choose| because it's
a common notation for the so-called ^{binomial coefficient}
that tells how many ways there are to choose $k$~things out of $n$~things.

You can't mix |\over| and |\atop| and |\choose| with each other.
For example, `|$$n \choose k \over 2$$|' is illegal; you must use
grouping, to get either `|$${n\choose k}\over2$$|' or
`|$$n\choose{k\over2}$$|', i.e.,
\begindisplay
$\displaystyle{{n\choose k}\over2}\qquad{\rm or}\qquad {n\choose{k\over2}}.$
\enddisplay
The latter formula, incidentally, would look better as
`|$$n\choose k/2$$|' or `|$$n\choose{1\over2}k$$|', yielding
\begindisplay
$\displaystyle{n\choose k/2}\qquad{\rm or}\qquad{n\choose{1\over2}k}.$
\enddisplay

\medskip
\exercise As alternatives to $\displaystyle{{n\choose k}\over2}$,
discuss how you could obtain the two displays
\begindisplay\abovedisplayskip=0pt\belowdisplayskip=0pt
$\displaystyle
{1\over2}{n\choose k}
\qquad{\rm and}\qquad
{\displaystyle{n\choose k}\over2}.$
\enddisplay
\answer |$${1\over2}{n\choose k}$$|;
|$$\displaystyle{n\choose k}\over2$$|.
All of these braces are necessary.

\bigbreak
\exercise Explain how to specify the displayed formula
$${p \choose 2}x^2 y^{p-2} - {1 \over 1-x}{1 \over 1-x^2}.$$
\answer |$${p \choose 2} x^2 y^{p-2} - {1 \over 1-x}{1 \over 1-x^2}.$$|

\danger \TeX\ has a generalized version of\/ |\over| and |\atop| in which you
specify the exact thickness of the line rule by typing
`^|\above|\<dimen>'. For example,
\begintt
$$\displaystyle{a\over b}\above1pt\displaystyle{c\over d}$$
\endtt
will produce a ^{compound fraction} with a heavier ($1\pt$ thick) rule as
its main bar:
$${\displaystyle{a\over b}\above 1pt\displaystyle{c\over d}}.$$
This sort of thing occurs primarily in textbooks on elementary mathematics.

\goodbreak
Mathematicians often use the sign $\sum$ to stand for ``^{summation}''
and the sign $\int$ to stand for ``^{integration}.'' If you're a typist but not
a mathematician, all you need to remember is that ^|\sum| stands for
$\sum$ and ^|\int| for $\int$; these abbreviations appear in Appendix~F
together with all the other symbols, in case you forget. Symbols like
$\sum$ and $\int$ (and a few others like $\bigcup$ and $\prod$ and $\oint$
and~$\bigotimes$, all listed in Appendix~F) are called {\sl ^{large operators}},
^^{collective signs, see large operators} ^^{sigma signs, see sum}
and you type them just as you type ordinary symbols or letters. The
difference is that \TeX\ will choose a {\sl larger\/} large operator in
display style than it will in text style. For example,
$$\halign{\indent#\hfil\qquad yields\qquad&$#\hfil$\qquad&#\hfil\cr
|$\sum x_n$|&\sum x_n&($T$ style)\cr
\noalign{\vskip3pt}
|$$\sum x_n$$|&\displaystyle\sum x_n&($D$ style).\cr}$$

A displayed |\sum| usually occurs with ``^{limits},'' i.e., with
subformulas that are to appear above and below it. You type limits just
as if they were superscripts and subscripts; for example, if you want
$$\sum_{n=1}^m$$
you type either `|$$\sum_{n=1}^m$$|' or `|$$\sum^m_{n=1}$$|'.  According
to the normal conventions of mathematical typesetting, \TeX\ will change
this to `$\sum_{n=1}^m$' (i.e., without limits) if it occurs in text
style rather than in display style.

Integrations are slightly different from summations, in that the superscripts
and subscripts are not set as limits even in display style:
$$\halign{\indent\hbox to2.3in{#\hfil}\hbox to.6in{yields\hfil}&
  $#\hfil$\qquad&#\hfil\cr
|$\int_{-\infty}^{+\infty}$|&\int_{-\infty}^{+\infty}&($T$ style)\cr
\noalign{\vskip3pt}
|$$\int_{-\infty}^{+\infty}$$|&\displaystyle\int_{-\infty}^{+\infty}&
  ($D$ style).\cr}$$

\danger Some printers prefer to set limits above and below $\int$ signs;
this takes more space on the page, but it
gives a better appearance if the subformulas are complex, because it
keeps them out of the way of the rest of the formula. Similarly, limits
are occasionally desirable in text style or script style; but some
printers prefer not to set limits on displayed $\sum$ signs. You can change
\TeX's convention by simply typing `^|\limits|' or `^|\nolimits|' immediately
after the large operator.
For example,
$$\halign{\indent\hbox to2.3in{#\hfil}\hbox to.6in{yields\hfil}&
  $\displaystyle{#}$\hfil\cr
|$$\int\limits_0^{\pi\over2}$$|&\int\limits_0^{\pi\over2}\cr
\noalign{\vskip 4pt}
|$$\sum\nolimits_{n=1}^m$$|&\sum\nolimits_{n=1}^m\cr}$$

\ddanger If you say `|\nolimits\limits|' (presumably because some macro
like |\int| specifies |\nolimits|, but you do want them), the last word
takes precedence.  There's also a command `^|\displaylimits|' that can be
used to restore \TeX's normal conventions; i.e., the limits will be
displayed only in styles $D$ and $D'$.

\danger Sometimes you need to put two or more rows of limits under a large
operator; you can do this with `^|\atop|'. For example, if you want
the displayed formula
$$\sum_{\scriptstyle0\le i\le m\atop\scriptstyle0<j<n}P(i,j)$$
the correct way to type it is
\begintt
$$\sum_{\scriptstyle0\le i\le m\atop\scriptstyle0<j<n}P(i,j)$$
\endtt
(perhaps with a few more spaces to make it look nicer in the manuscript
file). The instruction `^|\scriptstyle|' was necessary here,
twice---otherwise the lines `$0\le i\le m$' and `$0<j<n$' would have been in
scriptscript size, which is too small. This is another instance of a rare
case where \TeX's automatic style rules need to be overruled.

\exercise How would you type the displayed formula $\displaystyle
\sum_{i=1}^p\sum_{j=1}^q\sum_{k=1}^ra_{ij}b_{jk}c_{ki}$\enspace?
\answer |$$\sum_{i=1}^p\sum_{j=1}^q\sum_{k=1}^ra_{ij}b_{jk}c_{ki}$$|.

\dangerexercise And how would you handle $\displaystyle
\sum_{{\scriptstyle1\le i\le p\atop\scriptstyle1\le j\le q}
    \atop\scriptstyle1\le k\le r}a_{ij}b_{jk}c_{ki}$\enspace?
\answer |$$\sum_{{\scriptstyle 1\le i\le p \atop \scriptstyle 1\le j\le q}
    \atop \scriptstyle 1\le k\le r} a_{ij} b_{jk} c_{ki}$$|.

Since mathematical formulas can get horribly large, \TeX\ has to have some
way to make ever-larger symbols. For example, if you type
\begintt
$$\sqrt{1+\sqrt{1+\sqrt{1+
            \sqrt{1+\sqrt{1+\sqrt{1+\sqrt{1+x}}}}}}}$$
\endtt
the result shows a variety of available ^{square-root signs}:
\begindisplay
$\displaystyle\sqrt{1+\sqrt{1+\sqrt{1+
            \sqrt{1+\sqrt{1+\sqrt{1+\sqrt{1+x}}}}}}}$
\enddisplay
The three largest signs here are all essentially the same, except for a
vertical segment `\vbox{\hbox{\tenex\char'165}\vss}' that gets repeated as
often as necessary to reach the desired size; but the smaller signs are
distinct characters found in \TeX's math fonts.

A similar thing happens with parentheses and other so-called
``^{delimiter}'' symbols.  For example, here are some of the different sizes of
^^{fences, see delimiters}
^{parentheses} and ^{braces} that plain \TeX\ might use in formulas:
\begindisplay
$\displaystyle
\left(\vbox to 27pt{}\left(\vbox to 24pt{}\left(\vbox to 21pt{}
\Biggl(\biggl(\Bigl(\bigl(({\scriptstyle({\scriptscriptstyle(\hskip3pt
)})})\bigr)\Bigr)\biggr)\Biggr)\right)\right)\right)
\left\{\vbox to 27pt{}\left\{\vbox to 24pt{}\left\{\vbox to 21pt{}
\Biggl\{\biggl\{\Bigl\{\bigl\{\{{\scriptstyle\{{\scriptscriptstyle\{\hskip3pt
\}}\}}\}\bigr\}\Bigr\}\biggr\}\Biggr\}\right\}\right\}\right\}$
\enddisplay
The three largest pairs in each case are made with repeatable extensions,
so they can become as large as necessary.
^^{pieces of symbols}

Delimiters are important to mathematicians, because they provide good
visual clues to the underlying structure of complex expressions; they delimit
the boundaries of individual subformulas. Here is a list of the 22~basic
delimiters provided by plain \TeX:
\begindisplay
\it Input&\it Delimiter\cr
\noalign{\vskip2pt}
|(|&left parenthesis: $($\cr
|)|&right parenthesis: $)$\cr
|[| or ^|\lbrack|&left bracket: $[$\cr
|]| or ^|\rbrack|&right bracket: $]$\cr
|\{| or ^|\lbrace|&left curly brace: $\{$\cr
|\}| or ^|\rbrace|&right curly brace: $\}$\cr
^|\lfloor|&left floor bracket: $\lfloor$\cr
^|\rfloor|&right floor bracket: $\rfloor$\cr
^|\lceil|&left ceiling bracket: $\lceil$\cr
^|\rceil|&right ceiling bracket: $\rceil$\cr
^|\langle|&left angle bracket: $\langle$\cr
^|\rangle|&right angle bracket: $\rangle$\cr
|/|&slash: $/$\cr
^|\backslash|&reverse slash: $\backslash$\cr
\| or ^|\vert|&vertical bar: $\vert$\cr
|\|\| or ^|\Vert|&double vertical bar: $\Vert$\cr
^|\uparrow|&upward arrow: $\uparrow$\cr
^|\Uparrow|&double upward arrow: $\Uparrow$\cr
^|\downarrow|&downward arrow: $\downarrow$\cr
^|\Downarrow|&double downward arrow: $\Downarrow$\cr
^|\updownarrow|&up-and-down arrow: $\updownarrow$\cr
^|\Updownarrow|&double up-and-down arrow: $\Updownarrow$\cr
\enddisplay
^^{bent bars, see langle, rangle} ^^{curly braces, see lbrace, rbrace}
^^{leftbracket}^^{rightbracket}^^{leftbrace}^^{rightbrace}^^{/}
In some cases, there are two ways to get the same delimiter; for example,
you can specify a left bracket by typing either `|[|' or `|\lbrack|'. The
latter alternative has been provided because the symbol `|[|' is not
readily available on all computer keyboards. Remember, however,
that you should never try to specify a left brace or right brace simply by
typing `|{|' or `|}|'; the |{| and |}| symbols are reserved for grouping.
The right way is to type `|\{|' or `|\}|' or `|\lbrace|' or `|\rbrace|'.

In order to get a slightly larger version of any of these symbols, just
precede them by `^|\bigl|' (for opening delimiters) or `^|\bigr|' (for
closing ones). This makes it easier to read formulas that contain
delimiters inside delimiters:
\beginlongmathdemo
\it Input&\it Output\cr
\noalign{\vskip2pt}
|$\bigl(x-s(x)\bigr)\bigl(y-s(y)\bigr)$|&
  \bigl(x-s(x)\bigr)\bigl(y-s(y)\bigr)\cr
|$\bigl[x-s[x]\bigr]\bigl[y-s[y]\bigr]$|&
  \bigl[x-s[x]\bigr]\bigl[y-s[y]\bigr]\cr
|$\bigl|\|| |\||x|\||-|\||y|\|| \bigr|\||$|&
  \bigl\vert\vert x\vert-\vert y\vert\bigr\vert\cr
|$\bigl\lfloor\sqrt A\bigr\rfloor$|&
  \bigl\lfloor\sqrt A\bigr\rfloor\cr
\endmathdemo
The |\big| delimiters are just enough bigger than ordinary ones so that
the difference can be perceived, yet small enough to be used in the text
of a paragraph. Here are all~22 of them, in the ordinary size and in
the |\big| size:
\begindisplay
$(\,)\,[\,]\,\{\,\}\,\lfloor\,\rfloor\,\lceil\,\rceil\,\langle\,\rangle
  \,/\,\backslash\,\vert\,\Vert\,\uparrow\,\Uparrow\,\downarrow\,\Downarrow
  \,\updownarrow\,\Updownarrow$\cr
\noalign{\smallskip}
$\bigl(\,\bigr)\,\bigl[\,\bigr]\,\bigl\{\,\bigr\}\,\bigl\lfloor
  \,\bigr\rfloor\,\bigl\lceil\,\bigr\rceil\,\bigl\langle\,\bigr\rangle
  \,\big/\,\big\backslash\,\big\vert\,\big\Vert\,\bigm\uparrow\,\bigm\Uparrow
  \,\bigm\downarrow\,\bigm\Downarrow\,\bigm\updownarrow\,\bigm\Updownarrow$\cr
\enddisplay
You can also type ^|\Bigl| and ^|\Bigr| to get larger symbols suitable for
displays:
\begindisplay
$\Bigl(\,\Bigr)\,\Bigl[\,\Bigr]\,\Bigl\{\,\Bigr\}\,\Bigl\lfloor
  \,\Bigr\rfloor\,\Bigl\lceil\,\Bigr\rceil\,\Bigl\langle\,\Bigr\rangle
  \,\Big/\,\Big\backslash\,\Big\vert\,\Big\Vert\,\Bigm\uparrow\,\Bigm\Uparrow
  \,\Bigm\downarrow\,\Bigm\Downarrow\,\Bigm\updownarrow\,\Bigm\Updownarrow$
\enddisplay
These are 50\% taller than their |\big| counterparts. Displayed formulas
most often use delimiters that are even taller (twice the size of\/ |\big|);
such delimiters are constructed by ^|\biggl| and ^|\biggr|, and they
look like this:
\begindisplay
$\biggl(\,\biggr)\,\biggl[\,\biggr]\,\biggl\{\,\biggr\}\,\biggl\lfloor
  \,\biggr\rfloor\,\biggl\lceil\,\biggr\rceil\,\biggl\langle\,\biggr\rangle
  \,\bigg/\,\bigg\backslash\,\bigg\vert\,\bigg\Vert\,\biggm\uparrow
  \,\biggm\Uparrow\,\biggm\downarrow\,\biggm\Downarrow\,\biggm\updownarrow
  \,\biggm\Updownarrow$
\enddisplay
Finally, there are ^|\Biggl| and ^|\Biggr| versions, 2.5 times as tall
as the |\bigl| and |\bigr| delimiters:
\begindisplay
$\Biggl(\,\Biggr)\,\Biggl[\,\Biggr]\,\Biggl\{\,\Biggr\}\,\Biggl\lfloor
  \,\Biggr\rfloor\,\Biggl\lceil\,\Biggr\rceil\,\Biggl\langle\,\Biggr\rangle
  \,\Bigg/\,\Bigg\backslash\,\Bigg\vert\,\Bigg\Vert\,\Biggm\uparrow
  \,\Biggm\Uparrow\,\Biggm\downarrow\,\Biggm\Downarrow\,\Biggm\updownarrow
  \,\Biggm\Updownarrow$
\enddisplay

\medskip
\exercise Guess how to type the formula $\displaystyle
\biggl({\partial^2\over\partial x^2}+{\partial^2\over\partial y^2}
  \biggr)\bigl\vert\varphi(x+iy)\bigr\vert^2=0$, in display style,
using |\bigg| delimiters for the large parentheses. \ (The symbols $\partial$
and $\varphi$ that appear here are called ^|\partial| and ^|\varphi|.)
\answer |$\displaystyle\biggl({\partial^2\over\partial x^2}+|\hfil\break
|{\partial^2\over\partial y^2}\biggr)\bigl|\||\varphi(x+iy)\bigr|\||^2=0$|.

\dangerexercise In practice, |\big| and |\bigg| delimiters are used much
more often than |\Big| and |\Bigg| ones. Why do you think this is true?
\answer Formulas that are more than one line tall are usually two lines tall,
not 1$1\over2$ or 2$1\over2$ lines tall.

\danger A |\bigl| or |\Bigl| or |\biggl| or |\Biggl| delimiter is an
^{opening}, like a left parenthesis;
a |\bigr| or |\Bigr| or |\biggr| or |\Biggr| delimiter is a
^{closing}, like a right parenthesis. Plain \TeX\ also provides
^|\bigm| and ^|\Bigm| and ^|\biggm| and ^|\Biggm| delimiters, for use
in the middle of formulas; such a delimiter plays the r\^ole of a ^{relation},
like an equals sign, so \TeX\ puts a bit of space on either side of it.
\beginlongmathdemo
|$\bigl(x\in A(n)\bigm|\||x\in B(n)\bigr)$|&
  \tenmath\bigl(x\in A(n)\bigm\vert x\in B(n)\bigr)\cr
\noalign{\vskip2pt}
|$\bigcup_n X_n\bigm\|\||\bigcap_n Y_n$|&
  \tenmath\bigcup_n X_n\bigm\Vert\bigcap_n Y_n\cr
\endmathdemo
^^|\bigcup|^^|\bigcap|^^|\verticalline|^^|\in|
You can also say just ^|\big| or ^|\Big| or ^|\bigg| or ^|\Bigg|; this produces
a delimiter that acts as an ordinary variable. It is used primarily with
slashes and backslashes, as in the following example.
\beginlongmathdemo
\noalign{\vskip-2pt}
|$${a+1\over b}\bigg/{c+1\over d}$$|&
  \tenmath\displaystyle{a+1\over b}\bigg/{c+1\over d}\cr
\endmathdemo

\dangerexercise What's the professional way to type
$\tenmath\bigl(x+f(x)\bigr)\big/\bigl(x-f(x)\bigr)$? \ (Look closely.)
\answer |$\bigl(x+f(x)\bigr) \big/ \bigl(x-f(x)\bigr)$|. \ Notice especially
the `|\big/|'; an ordinary ^{slash} would look too small between the
|\big| parentheses.

\TeX\ has a built-in mechanism that figures out how tall a pair of delimiters
needs to be, in order to enclose a given subformula; so you can use this
method, instead of deciding whether a delimiter should be |\big| or
|\bigg| or whatever. All you do is say
\begindisplay
^|\left|\<delim$_1$>\<subformula>^|\right|\<delim$_2$>
\enddisplay
and \TeX\ will typeset the subformula, putting the specified delimiters at
the left and the right. The size of the delimiters will be just big enough
to cover the subformula. For example, in the display
\beginlongdisplaymathdemo
|$$1+\left(1\over1-x^2\right)^3$$|&1+\left(1\over1-x^2\right)^3\cr
\endmathdemo
\TeX\ has chosen |\biggl(| and |\biggr)|, because smaller delimiters
would be too small for this particular fraction. A simple formula like
`|$\left(x\right)$|' yields just `$\left(x\right)$'; thus, |\left| and
|\right| sometimes choose delimiters that are smaller than |\bigl| and |\bigr|.

Whenever you use |\left| and |\right| they must pair up with each other,
just as braces do in groups. You can't have |\left| in one formula
and |\right| in another, nor are you allowed to type things like
`|\left(...{...\right)...}|' or
`|\left(...\begingroup...\right)...\endgroup|'.
This restriction makes sense, because \TeX\ needs to typeset the
subformula that appears between |\left| and |\right| before it can decide
how big to make the delimiters.  But it is worth explicit mention here,
because you do {\sl not\/} have to match ^{parentheses} and ^{brackets}, etc.,
^^{crotchets, see brackets}
when you are not using |\left| and |\right|: \TeX\ will not complain if
you input a formula like `|$[0,1)$|' or even `|$)($|' or just `|$)$|'.\
(And it's a good thing \TeX\ doesn't, for such unbalanced formulas occur
surprisingly often in mathematics papers.) \ Even when you do use |\left|
and |\right|, \TeX\ doesn't look closely at the particular delimiters that
you happen to choose; thus, you can type strange things like `|\left)|'
and/or `|\right(|' if you know what you're doing. Or even if you don't.

The |\over| operation in the example displayed above does not involve the
`|1+|' at the beginning of the formula; this happens because |\left| and
|\right| have the function of ^{grouping}, in addition to their function
of delimiter-making.  Any definitions that you happen to make between
|\left| and |\right| will be local, as if braces had appeared around the
enclosed subformula.

\exercise Use |\left| and |\right| to typeset the following display
(with ^|\phi| for $\phi$):
$$\pi(n)=\sum_{k=2}^n\left\lfloor\phi(k)\over k-1\right\rfloor.$$
\answer |$$\pi(n)=\sum_{k=2}^n\left\lfloor\phi(k)\over k-1\right\rfloor.$$|

At this point you are probably wondering why you should bother learning about
|\bigl| and |\bigr| and their relatives, when |\left| and |\right| are there
to calculate sizes for you automatically. Well, it's true that |\left|
and |\right| are quite handy, but there are at least three situations in which
you will want to use your own wisdom when selecting the proper delimiter size:
\ (1)~Sometimes |\left| and |\right| choose a smaller delimiter than you want.
For example, we used |\bigl| and |\bigr| to produce $\bigl\vert\vert x\vert-
\vert y\vert\bigr\vert$ in one of the previous illustrations; |\left| and
|\right| don't make things any bigger than necessary, so
`|$\left|\||\left|\||x\right|\||-\left|\||y\right|\||\right|\||$|'
yields only
`$\left\vert \left\vert x\right\vert -\left\vert y\right\vert \right\vert$'.
\ (2)~Sometimes |\left| and |\right| choose a larger delimiter than you want.
This happens most frequently when they enclose a large operator in a display;
for example, compare the following two formulas:
\beginlongdisplaymathdemo
\noalign{\vskip 6pt}
|$$\left( \sum_{k=1}^n A_k \right)$$|&\left( \sum_{k=1}^n A_k \right)\cr
\noalign{\vskip 3pt}
|$$\biggl( \sum_{k=1}^n A_k \biggr)$$|&\biggl( \sum_{k=1}^n A_k \biggr)\cr
\endmathdemo
The rules of\/ |\left| and |\right| cause them to enclose the ^|\sum| together
with its ^{limits}, but in special cases like this it looks better to let
the limits hang out a~bit; |\bigg| delimiters are better here.
\ (3)~Sometimes you need to break a huge displayed
formula into two or more separate lines, and you want to make sure that
its opening and closing delimiters have the same size; but you can't use
|\left| on the first line and |\right| on the last, since |\left| and
|\right| must occur in pairs. The solution is to use |\Biggl| (say) on
the first line and |\Biggr| on the last.

\danger Of course, one of the advantages of\/ |\left| and |\right| is that
they can make arbitrarily large delimiters---much bigger than |\biggggg|!
The slashes and angle brackets do have a maximum size, however; if you
ask for really big versions of those symbols you will get the largest
ones available.

\exercise Prove that you have mastered delimiters: Coerce \TeX\ into
producing the formula
$$\pi(n)=\sum_{m=2}^n\left\lfloor\biggl(\sum_{k=1}^{m-1}\bigl\lfloor
  (m/k)\big/\lceil m/k\rceil\bigr\rfloor\biggr)^{-1}\right\rfloor.$$
\answer |$$\pi(n)=\sum_{m=2}^n\left\lfloor\biggl(\sum_{k=1}^{m-1}\bigl|
\hfil\break
|\lfloor(m/k)\big/\lceil m/k\rceil\bigr\rfloor\biggr)^{-1}\right\rfloor.$$|

\danger If you type `|.|'\ after |\left| or |\right|, instead of
specifying one of the basic delimiters, you get a so-called ^{null
delimiter} (which is blank). Why on earth would anybody want that, you may
ask. Well, you sometimes need to produce formulas that contain only one
large delimiter. For example, the display
$$\vert x\vert=\cases{x,&if $x\ge0$\cr
  -x,&if $x<0$\cr}$$
has a `$\{$' but no `$\}$'. It can be produced by a construction of the form
\begindisplay
|$$|\||x|\||=\left\{ ... \right.$$|
\enddisplay
Chapter 18 explains how to fill in the `\hbox{|...|}' to finish this
construction; let's just notice for now that the `|\right.|'\ makes it
possible to have an invisible right delimiter to go with the visible
left brace.

\ddanger A null delimiter isn't completely void; it is an empty box
whose width is a \TeX\ parameter called ^|\nulldelimiterspace|.
We will see later that null delimiters are inserted next to fractions.
Plain \TeX\ sets |\nulldelimiterspace=1.2pt|.

You can type `|<|' or `|>|' as convenient abbreviations for ^|\langle| and
^|\rangle|, when \TeX\ is looking for a delimiter. For example,
`|\bigl<|' is equivalent to `|\bigl\langle|', and `|\right>|' is
equivalent to `|\right\rangle|'. Of course `|<|' and `|>|' ordinarily
produce the ^{less-than} and ^{greater-than} relations `${<}\,{>}$', which
are quite different from ^{angle brackets} `$\langle\,\rangle$'.

\danger Plain \TeX\ also makes available a few more delimiters, which were
not listed in the basic set of~22 because they are sort of special.
The control sequences ^|\arrowvert|, ^|\Arrowvert|, and ^|\bracevert| produce
delimiters made from the repeatable parts of the vertical arrows, double
vertical arrows, and large braces, respectively, without the arrowheads
or the curly parts of the braces. They produce results similar to
^|\vert| or ^|\Vert|, but they are surrounded by more white space and
they have a different weight. You can also use ^|\lgroup| and ^|\rgroup|,
which are constructed from braces without the middle parts; and
^|\lmoustache| and ^|\rmoustache|, ^^{moustaches}
which give you the top and bottom halves of large braces. For example,
here are the |\Big| and |\bigg| versions of\/ |\vert|, |\Vert|, and
these seven special delimiters:
$$\halign{\indent$#\hfil$\cr
\ldots\Big\vert\ldots\Big\Vert
\ldots\Big\arrowvert\ldots\Big\Arrowvert\ldots\Big\bracevert
\ldots\Big\lgroup\ldots\Big\rgroup\ldots\Big\lmoustache\ldots\Big\rmoustache
\ldots\,;\cr
\noalign{\smallskip}
\ldots\bigg\vert\ldots\bigg\Vert
\ldots\bigg\arrowvert\ldots\bigg\Arrowvert\ldots\bigg\bracevert
\ldots\bigg\lgroup\ldots\bigg\rgroup\ldots\bigg\lmoustache\ldots\bigg\rmoustache
\ldots\,.\cr}$$
Notice that |\lgroup| and |\rgroup| are rather like bold parentheses, with
sharper bends at the corners; this makes them attractive for certain large
displays. But you cannot use them exactly like parentheses, because
they are available only in large sizes (|\Big|~or~more).

\ddanger Question: What happens if a ^{subscript} or ^{superscript}
follows a large delimiter? Answer:~That's a good question. After a |\left|
delimiter, it is the first subscript or superscript of the enclosed
subformula, so it is effectively preceded by |{}|. After a |\right|
delimiter, it is a subscript or superscript of the entire |\left...\right|
subformula. And after a |\bigl| or |\bigr| or |\bigm| or |\big| delimiter,
it applies only to that particular delimiter. Thus, `|\bigl(_2|' works
quite differently from `|\left(_2|'.

\danger If you look closely at the examples of math typesetting in this
chapter, you will notice that large parentheses and brackets are
symmetric with respect to an invisible horizontal line that runs a little
bit above the ^{baseline}; when a delimiter gets larger, its height and
depth both grow by the same amount. This horizontal line is called the
{\sl^{axis}\/} of the formula; for example, a formula in the text of the
present paragraph would have an axis at this level: $\hskip 2em\over$. The
bar line in every fraction is centered on the axis, regardless of the size
of the numerator or denominator.

\danger Sometimes it is necessary to create a special box that should be
centered vertically with respect to the axis. \ (For example, the
`$\vert x\vert=\bigl\{\,\ldots$' example above was done with such a box.) \
\TeX\ provides a simple way to do this: You just say
\begindisplay
|\vcenter{|\<vertical mode material>|}|
\enddisplay
and the vertical mode material will be packed into a box just as if
^|\vcenter| had been ^|\vbox|. Then the box will be raised or lowered until
its top edge is as far above the axis as the bottom edge is below.

\ddanger The concept of ``axis'' is meaningful for \TeX\ only in math
formulas, not in ordinary text; therefore \TeX\ allows you to use
|\vcenter| only in math mode. If you really need to center something
vertically in horizontal mode, the solution is to say `|$\vcenter{...}$|'.
 \ (Incidentally, the constructions `|\vcenter| |to|\<dimen>'
and `|\vcenter| |spread|\<dimen>' are legal too, in math mode;
vertical glue is always set by the rules for |\vbox| in
Chapter~12. But |\vcenter| by itself is usually sufficient.)

\danger Any box can be put into a formula by simply saying ^|\hbox| or
|\vbox| or ^|\vtop| or ^|\box| or ^|\copy| in the normal way, even when
you are in math mode.  Furthermore you can use ^|\raise| or ^|\lower|, as
if you were in horizontal mode, and you can insert vertical rules with
^|\vrule|.  Such constructions, like |\vcenter|, produce boxes that can be
used like ordinary symbols in math formulas.

\ddanger Sometimes you need to make up your own symbols, when you run across
something unusual that doesn't occur in the fonts. If the new symbol
occurs only in one place, you can use |\hbox| or |\vcenter| or something
to insert exactly what you want; but if you are defining a macro for
general use, you may want to use different constructions in different
styles. \TeX\ has a special feature called ^|\mathchoice| that comes
to the rescue in such situations: You write
\begindisplay
|\mathchoice{|\<math>|}{|\<math>|}{|\<math>|}{|\<math>|}|
\enddisplay
where each \<math> specifies a subformula. \TeX\ will choose the first
subformula in style $D$ or~$D'$, the second in style $T$ or~$T'$, the
third in style $S$ or~$S'$, the fourth in style $\SS$ or $\SS'$.
\ (\TeX\ actually typesets all four subformulas, before it chooses the
final one, because the actual style is not always known at the time a
|\mathchoice| is encountered; for example, when you type `|\over|' you often
change the style of everything that has occurred earlier in the formula.
Therefore |\mathchoice| is somewhat expensive in terms of time and space,
and you should use it only when you're willing to pay the price.)

\ddangerexercise Guess what output is produced by the following commands:
\begintt
\def\puzzle{{\mathchoice{D}{T}{S}{SS}}}
$$\puzzle{\puzzle\over\puzzle^{\puzzle^\puzzle}}$$
\endtt
\answer A displayed formula equivalent to |$${D}{{T}\over{T}^{{S}^{SS}}}$$|.

\ddangerexercise Devise a `^|\square|' macro that produces a
\def\sqr#1#2{{\vcenter{\vbox{\hrule height.#2pt
      \hbox{\vrule width.#2pt height#1pt \kern#1pt \vrule}
      \hrule height.#2pt}}}}%
`$\,\sqr34\,$' for use in math formulas. The box should be symmetrical
with respect to the axis, and its inside dimensions should be $3\pt$ in
display and text styles, $2.1\pt$ in script styles, and $1.5\pt$ in
scriptscript styles. The rules should be $0.4\pt$ thick in
display and text styles, $0.3\pt$ thick otherwise.
\answer |\def\sqr#1#2{{\vcenter{\vbox{\hrule height.#2pt|\parbreak
        |        \hbox{\vrule width.#2pt height#1pt \kern#1pt|\parbreak
        |           \vrule width.#2pt}|\parbreak
        |        \hrule height.#2pt}}}}|\parbreak
        |\def\square{\mathchoice\sqr34\sqr34\sqr{2.1}3\sqr{1.5}3}|

\ddanger Plain \TeX\ has a macro called ^|\mathpalette| that is useful
for |\mathchoice| constructions; `|\mathpalette\a{xyz}|' expands to
the four-pronged array of choices
`|\mathchoice|\stretch|{\a|\stretch|\displaystyle|\stretch|{xyz}}|\stretch
|...|\stretch|{\a|\stretch|\scriptscriptstyle|\stretch|{xyz}}|\stretch'.
Thus the first argument to |\mathpalette| is a control sequence whose
first argument is a style selection. Appendix~B contains several examples
that show how |\mathpalette| can be applied. \ (See in particular the
definitions of\/ |\phantom|, |\root|, and |\smash|; the ^{congruence sign}
^|\cong| ($\cong$) is also constructed from $=$ and $\sim$ using
|\mathpalette|.)
^^{constructing new math symbols}
^^{math symbols, construction of}

\ddanger At the beginning of this chapter we discussed the commands
|\over|, |\atop|, |\choose|, and |\above|. These are special cases of
\TeX's ``^{generalized fraction}'' feature, which includes also the
three primitives
\begindisplay
|\overwithdelims|\<delim$_1$>\<delim$_2$>\cr
|\atopwithdelims|\<delim$_1$>\<delim$_2$>\cr
|\abovewithdelims|\<delim$_1$>\<delim$_2$>\<dimen>\cr
\enddisplay
The third of these is the most general, as it encompasses all of the other
generalized fractions:  ^|\overwithdelims| uses a ^{fraction} bar whose
thickness is the default for the current size, and ^|\atopwithdelims| uses
an invisible fraction bar whose thickness is zero, while
^|\abovewithdelims| uses a bar whose thickness is specified explicitly.
\TeX\ places the immediately preceding subformula (the ^{numerator}) over
the immediately following subformula (the ^{denominator}), separated by a
bar line of the desired thickness; then it puts \<delim$_1$> at the left
and \<delim$_2$> at the right. For example, `^|\choose|' is equivalent to
`|\atopwithdelims()|'. If you define |\legendre| to be
`|\overwithdelims()|', you can typeset the ^{Legendre symbol}
\def\legendre{\overwithdelims()}%
`$a\legendre b$' by saying `|{a\legendre b}|'. The size of the surrounding
delimiters depends only on the style, not on the size of the fractions;
larger delimiters are used in styles $D$ and~$D'$ (see Appendix~G\null). The
simple commands ^|\over|, ^|\atop|, and ^|\above| are equivalent to the
corresponding `|withdelims|' commands when the delimiters are null; for
example, `|\over|' is an abbreviation for `|\overwithdelims..|'.

\def\euler{\atopwithdelims<>}
\ddangerexercise Define a control sequence |\euler| so that the
^{Eulerian number} $n\euler k$ will be produced when you type `|{n\euler k}|'
in a formula.
\answer|\def\euler{\atopwithdelims<>}|.

\ddanger Appendix G explains exactly how \TeX\ computes the desired size
of delimiters for |\left| and~|\right|. The general idea is that delimiters
are vertically centered with respect to the ^{axis}; hence, if we want
to cover a subformula between |\left| and |\right| that extends $y_1$~units
above the axis and $y_2$~units below, we need to make a delimiter whose
height plus depth is at least $y$~units, where $y=2\max(y_1,y_2)$.
It is usually best not to cover the formula completely, however,
but just to come close; so \TeX\ allows you to specify
two parameters, the ^|\delimiterfactor|~$f$ (an~integer) and the
^|\delimitershortfall|~$\delta$ (a~dimension). The minimum delimiter size
is taken to be at least $y\cdot f/1000$, and at least $y-\delta$. Appendix~B
sets $f=901$ and $\delta=5\pt$. Thus, if $y=30\pt$, the plain \TeX\ format
causes the delimiter to be more than $27\pt$ tall; if $y=100\pt$, the
corresponding delimiter will be at least $95\pt$ tall.

\danger So far we have been discussing the rules for typing math formulas,
but we haven't said much about how \TeX\ actually goes about converting
its input into lists of boxes and glue. Almost all of the control
sequences that have been mentioned in Chapters 16 and~17 are ``high level''
features of the plain \TeX\ format; they are not built into \TeX\ itself.
Appendix~B defines those control sequences in terms of more primitive
commands that \TeX\ actually deals with. For example, `|\choose|' is
an abbreviation for `|\atopwithdelims()|'; Appendix~B not only introduces
|\choose|, it also tells \TeX\ where to find the delimiters |(| and~|)|
in various sizes. The plain \TeX\ format defines all of the special
characters like |\alpha| and~|\mapsto|, all of the special accents like
|\tilde| and~|\widehat|, all of the large operators like |\sum| and~|\int|,
and all of the delimiters like |\lfloor| and~|\vert|. Any of these things
can be redefined, in order to adapt \TeX\ to other mathematical styles
and/or to other fonts.

\danger The remainder of this chapter discusses the low-level commands
that \TeX\ actually obeys behind the scenes.  Every paragraph on the next
few pages is marked with double dangerous bends, so you should skip to
Chapter~18 unless you are a glutton for \TeX nicalities.

\ninepoint
\ddanger All characters that are typeset in math mode belong to one of
sixteen {\sl^{families} of fonts}, numbered internally from 0 to~15. Each
of these families consists of three fonts: one for text size, one for
script size, and one for scriptscript size. The commands ^|\textfont|,
^|\scriptfont|, and ^|\scriptscriptfont| are used to specify the members
of each family.  For example, ^{family~0} in the plain \TeX\ format is
used for roman letters, and Appendix~B contains the instructions
\begintt
\textfont0=\tenrm
\scriptfont0=\sevenrm
\scriptscriptfont0=\fiverm
\endtt
to set up this family: The 10-point roman font (^|\tenrm|) is used for
normal symbols, 7-point roman (^|\sevenrm|) is used for subscripts, and
5-point roman (^|\fiverm|) is used for sub-subscripts. Since there are up to
256~characters per font, and 3~fonts per family, and 16~families, \TeX\ can
access up to 12,288 characters in any one formula (4096 in~each of
the three sizes). Imagine that.

\ddanger A definition like |\textfont|\<family number>|=|\<font identifier>
is local to the group that contains it, so you can easily change family
membership from one set of conventions to another and back again. Furthermore
you can put any font into any family; for example, the command
\begintt
\scriptscriptfont0=\scriptfont0
\endtt
makes sub-subscripts in family~0 the same size as the subscripts currently
are. \TeX\ doesn't check to see if the families are sensibly organized; it
just follows instructions. \ (However, fonts cannot be used in families
2 and~3 unless they contain a certain number of special parameters, as we
shall see later.) \ Incidentally, \TeX\ uses ^|\nullfont|, which contains
no characters, for each family member that has not been defined.

\ddanger During the time that a math formula is being read,
\TeX\ remembers each symbol as being ``character position so-and-so in
family number such-and-such,'' but it does not take note of what fonts
are actually in the families until reaching the end of the formula.
Thus, if you have loaded a font called |\Helvetica| that contains Swiss-style
numerals, and if you say something like
\begintt
$\textfont0=\tenrm 9 \textfont0=\Helvetica 9$
\endtt
you will get two 9's in font |\Helvetica|, assuming that \TeX\ has been
set up to take 9's from family~0. The reason is that |\textfont0|
is~|\Helvetica| at the end of the formula, and that's when it counts. On
the other hand, if you say
\begintt
$\textfont0=\tenrm 9 \hbox{$9\textfont0=\Helvetica$}$
\endtt
the first 9 will be from |\tenrm| and the second from |\Helvetica|, because
the formula in the hbox will be typeset before it is incorporated into
the surrounding formula.

\ddangerexercise If you say `|${\textfont0=\Helvetica 9}$|', what
font will be used for the~9?
\answer The |\textfont0| that was current at the beginning of the formula
will be used, because this redefinition is local to the braces. \
(It would be a different story if `^|\global||\textfont|' had appeared instead;
that would have changed the meaning of\/ |\textfont0| at all levels.)

\ddanger Every ^{math character} is given an identifying code
number between 0 and~4095, obtained by adding 256~times the family number
to the position number. This is easily expressed in ^{hexadecimal
notation}, using one hexadecimal digit for the family and two for the
character; for example, \hex{24A} stands for character~\hex{4A} in
family~2. Each character is also assigned to one of eight classes,
^^{classes of math characters, table} ^^{math codes} ^^{table of ...}
numbered 0 to~7, as follows:
$$\halign{\indent#\hfil&\quad#\hfil&\quad#\hfil&
\hskip4em#\hfil&\quad#\hfil&\quad#\hfil\cr
\it \kern-2pt Class&\it Meaning&\kern-2pt\it Example&
\it \kern-2pt Class&\it Meaning&\kern-2pt\it Example\cr
\noalign{\vskip2pt}
0&Ordinary&|/|&
4&Opening&|(|\cr
1&Large operator&|\sum|&
5&Closing&|)|\cr
2&Binary operation&|+|&
6&Punctuation&|,|\cr
3&Relation&|=|&
7&Variable family&|x|\cr
}$$
^^{large operator}^^{binary operation}^^{relation}^^{opening}^^{closing}
^^{punctuation}^^{variable family}
Classes 0 to 6 tell what ``part of speech'' the character belongs to, in
math-printing language; class~7 is a special case discussed below. The class
number is multiplied by 4096 and added to the character number, and this
is the same as making it the leading digit of a four-digit hexadecimal
number. For example, Appendix~B defines |\sum| to be the math character
\hex{1350}, meaning that it is a large operator (class~1) found in position
\hex{50} of family~3.

\ddangerexercise The ^|\oplus| and ^|\bullet| symbols ($\oplus$ and $\bullet$)
are binary operations that appear in positions 8 and~15 (decimal)
of family~2, when the fonts of plain~\TeX\ are being used. Guess
what their math character codes are. \ (This is too easy.)
\answer \hex{2208} and \hex{220F}.

\ddanger Class 7 is a special case that allows math symbols to change families.
It behaves exactly like class~0, except that the specified family is
replaced by the current value of an integer parameter called ^|\fam|,
provided that |\fam| is a legal family number (i.e., if it lies between
0 and~15). \TeX\ automatically sets |\fam=-1| whenever math mode is entered;
therefore class~7 and class~0 are equivalent unless |\fam| has been
given a new value. Plain \TeX\ changes |\fam| to~0 when the user
types `^|\rm|'; this makes it convenient to get roman letters in formulas,
as we will see in Chapter~18, since letters belong to class~7. \ (The
control sequence |\rm| is an abbreviation for `|\fam=0 \tenrm|'; thus,
|\rm| causes
|\fam| to become zero, and it makes |\tenrm| the ``^{current font}.''
In horizontal mode, the |\fam| value is irrelevant and the current font
governs the typesetting of letters; but in math mode, the current font is
irrelevant and the |\fam| value governs the letters. The current font
affects math mode only if\/ |\|\] is used ^^{control space} or if
dimensions are given in ^|ex| or ^|em| units;
it also has an effect if an |\hbox| appears inside a formula, since
the contents of an hbox are typeset in horizontal mode.)

\ddanger The interpretation of characters in math mode is defined by a
table of~256 ``mathcode'' values; these table entries can be changed
by the ^|\mathcode| command, just as the category codes are changed
by ^|\catcode| (see Chapter~7). Each mathcode specifies class, family, and
character position, as described above. For example, Appendix~B contains
the commands
\begintt
\mathcode`<="313C
\mathcode`*="2203
\endtt
which cause \TeX\ to treat the character `|<|' in math mode as a relation
^^{less than} (class~3) found in position \hex{3C} of family~1, and to treat an
^{asterisk} `|*|' as a binary operation found in position~3 of family~2.
The initial value of\/ |\mathcode`b| is \hex{7162}; thus, |b|~is character
\hex{62} in ^{family~1} (italics), and its family will vary with |\fam|.
\ (|INITEX| starts out with |\mathcode|$\,x=x$ for all characters~$x$
that are neither
letters nor digits. The ten digits have |\mathcode|$\,x=x+\hbox{\hex{7000}}$;
the 52 letters have |\mathcode|$\,x=x+\hbox{\hex{7100}}$.) \
\TeX\ looks at the mathcode only when it is typesetting a character whose
catcode is 11~(letter) or 12~(other), or when it encounters a character that
is given explicitly as ^|\char|\<number>.

\ddanger A |\mathcode| can also have the special value \hex{8000}, which
causes the character to behave as if it has catcode~13 (active). Appendix~B
uses this feature to make |'| ^^{apostrophe} expand to |^{|^|\prime||}| in a
slightly tricky way. The mathcode of |'| does not ^^{active math character}
interfere with the use of |'| in ^{octal} constants.

\ddanger The mathcode table allows you to refer indirectly to any character in
any family, with the touch of a single key. You can also specify a math
character code directly, by typing ^|\mathchar|, which is analogous to
^|\char|. For example, the command `|\mathchar"1ABC|' specifies a
character of class~1, family~10 (\hex A), and position \hex{BC}. A~hundred
or so definitions like
\begintt
\def\sum{\mathchar"1350 }
\endtt
would therefore suffice to define the special symbols of plain \TeX\null. But
there is a better way: \TeX\ has a primitive command ^|\mathchardef|,
which relates to |\mathchar| just as ^|\chardef| does to |\char|.
Appendix~B has a hundred or so definitions like
\begintt
\mathchardef\sum="1350
\endtt
to define the special symbols. A |\mathchar| must be between 0 and 32767
(\hex{7FFF}).

\ddanger A character of class~1, i.e., a ^{large operator} like |\sum|, will
be vertically centered with respect to the axis when it is typeset. Thus,
the large operators can be used with different sizes of type. This vertical
adjustment is not made for symbols of the other classes.

\ddanger \TeX\ associates classes with subformulas as well as with individual
characters. Thus, for example, you can treat a complex construction as if
it were a binary operation or a relation, etc., if you want to. The
commands ^|\mathord|, ^|\mathop|, ^|\mathbin|, ^|\mathrel|, ^|\mathopen|,
^|\mathclose|, and ^|\mathpunct| are used for this purpose; each of them
is followed either by a single character or by a subformula in braces.
For example, |\mathopen\mathchar"1234| is equivalent to |\mathchar"4234|,
because |\mathopen| forces class~4 (opening). In the formula
`|$G\mathbin:H$|', the ^{colon} is~treated as a binary operation.
And Appendix~B constructs large opening symbols by
defining ^|\bigl||#1| to be an abbreviation for
\begintt|let|0|rm
\mathopen{\hbox{$\left#1|<|0strut>\right.$}}
\endtt
There's also an eighth classification, ^|\mathinner|, which is not
normally used for individual symbols; ^|\left|$\,\ldots\,$^|\right|
constructions are treated as ``inner'' subformulas, which means that
they will be surrounded by additional space in certain circumstances.
All other subformulas are generally treated as ordinary symbols,
whether they are formed by |\overline| or |\hbox| or |\vcenter| or
by simply being enclosed in braces. Thus, |\mathord| isn't really
a necessary part of the \TeX\ language; instead of typing
`|$1\mathord,234$|' you can get the same effect from `|$1{,}234$|'.

\ddangerexercise Commands like |\mathchardef\alpha="010B| are used in
Appendix~B to define the lowercase ^{Greek} letters. Suppose that you want
to extend plain \TeX\ by putting ^{boldface math italic} letters
in family~9, analogous to the normal math italic letters in family~1.
\ (Such fonts aren't available in stripped down versions of \TeX, but
let's assume that they exist.) \ Assume that the control sequence
|\bmit| has been defined as an abbreviation for `|\fam=9|'; hence
`|{\bmit b}|' will give a boldface math italic~|b|. What change to the
definition of\/ |\alpha| will make |{\bmit\alpha}| produce a boldface~alpha?
\checkequals\bmiexno\exno
\answer |\mathchardef\alpha="710B|. Incidentally, |{\rm\alpha}| will
then give a spurious result, because character position \hex{0B} of
roman fonts does not contain an alpha; you should warn
your users about what characters they are allowed to type under the
influence of special conventions like ^|\rm|.

\ddanger ^{Delimiters} are specified in a similar but more complicated
way. Each character has not only a~|\catcode| and a~|\mathcode| but also
a~^|\delcode|, which is either negative (for characters that should not
act as delimiters) or less than \hex{1000000}. In other words,
nonnegative delcodes consist of six hexadecimal digits. The first three
digits specify a ``small'' variant of the delimiter, and the last three
specify a ``large'' variant. For example, the command
\begintt
\delcode`x="123456
\endtt
means that if the letter |x| is used as a delimiter, its small variant
is found in position \hex{23} of family~1, and its large variant is found
in position \hex{56} of family~4. If the small or large variant is
given as |000|, however (position~0 of ^{family~0}), that variant is ignored.
\TeX\ looks at the delcode when a character follows ^|\left| or ^|\right|,
or when a character follows one of the ^|withdelims| commands; a
negative delcode leads to an error message, but otherwise \TeX\ finds
a suitable delimiter by first trying the small variant and then
the large. \ (Appendix~G discusses this process in more detail.) \
For example, Appendix~B contains the commands
\begintt
\delcode`(="028300  \delcode`.=0
\endtt
which specify that the small variant of a left parenthesis is found in
position \hex{28} of family~0, and that the large variant is in position~0
of family~3; also, a period has no variants, hence `|\left.|'\ will produce
a ^{null delimiter}. There actually are several different left parenthesis
symbols in family~3; the smallest is in position~0, and the others are
linked together by information that comes with the font. All delcodes
are~$-1$ until they are changed by a |\delcode| command.

\ddangerexercise Appendix~B defines |\delcode`<| so that there is a
shorthand notation for ^{angle brackets}. Why do you think Appendix~B
doesn't go further and define |\delcode`{|?
\answer Assigning |\delcode`{| would not work to allow `|\left{|', because
the brace has category~1 and isn't a legal \<delim>.
Allowing brace delimiters would be a bad idea because it would
mess up other constructions, such as arguments to macros, and
components of alignments. Moreover, a user who
gets away with `|\left{|' is likely to try also `|\bigl{|', which
fails miserably.

\ddanger A delimiter can also be given directly, as `^|\delimiter|\<number>'.
In this case the number can be as high as \hex{7FFFFFF}, i.e., seven
hexadecimal digits; the leading digit specifies a class, from 0 to~7,
as in a |\mathchar|. For example, Appendix~B contains the definition
\begintt
\def\langle{\delimiter"426830A }
\endtt
and this means that ^|\langle| is an opening (class 4) whose small
variant is \hex{268} and whose large variant is \hex{30A}. When |\delimiter|
appears after |\left| or |\right|, the class digit is ignored; but
when |\delimiter| occurs in other contexts, i.e., when \TeX\ isn't
looking for a delimiter, the three rightmost digits are dropped and
the remaining four digits act as a |\mathchar|. For example, the expression
`|$\langle x$|' is treated as if it were `|$\mathchar"4268 x$|'.

\ddangerexercise What goes wrong if you type
`|\bigl\delimiter"426830A|'\thinspace?
\answer Since |\bigl| is defined as a macro with one parameter,
it gets just `|\delimiter|' as the argument. You have to write
`|\bigl{\delimiter"426830A}|' to make this work. On the other hand,
|\left| will balk if the following character is a left brace. Therefore
it's best to have control sequence names for all delimiters.

\ddanger Granted that these numeric conventions for |\mathchar| and
|\delimiter| are not beautiful, they sure do pack a lot of information into
a small space. That's why \TeX\ uses them for low-level definitions inside
formats. Two other low-level primitives also deserve to be mentioned:
^|\radical| and ^|\mathaccent|. Plain \TeX\ makes ^{square root signs}
and math accents available by giving the commands
\begintt
\def\sqrt{\radical"270370 }
\def\widehat{\mathaccent"362 }
\endtt
and several more like them. The idea is that |\radical| is followed by
a delimiter code and |\mathaccent| is followed by a math character code,
so that \TeX\ knows the family and character positions for the symbols
used in radical and accent constructions. Appendix~G gives precise
information about the positioning of these characters. By changing the
definitions, \TeX\ could easily be extended so that it would typeset a
variety of different radical signs and a variety of different accent
signs, if such symbols were available in the fonts.
^^{surd signs, see radical}

\ddanger Plain \TeX\ uses ^{family~1} for math italic letters, ^{family~2} for
ordinary math symbols, and ^{family~3} for large symbols. \TeX\ insists that
^^{math fonts}
the fonts in families 2 and~3 have special ^|\fontdimen| parameters,
which govern mathematical spacing according to the rules in Appendix~G\null;
the ^|cmsy| and ^|cmex| ^{symbol fonts} have these parameters, so
their assignment to families 2 and~3 is almost mandatory. \ (There is, however,
a way to modify the parameters of any font, using the ^|\fontdimen| command.) \
^|INITEX| initializes the mathcodes of all ^{letters} |A| to~|Z| and |a| to~|z|
so that they are symbols of class~7 and family~1; that's why
it is natural to use family~1 for math italics.  Similarly, the digits |0|
to~|9| are class~7 and family~0.  None of the other families
is treated in any special way by \TeX. Thus, for example, plain \TeX\ puts
^{text italic} in family~4, slanted roman in family~5, bold roman in family~6,
and typewriter type in family~7, but any of these numbers could be
switched around. There is a macro ^|\newfam|, analogous to |\newbox|,
that will assign symbolic names to families that aren't already used.

\ddanger When \TeX\ is in horizontal mode, it is making a horizontal list;
in vertical mode, it is making a vertical list. Therefore it should come
as no great surprise that \TeX\ is making a ^{math list} when it is in
^{math mode}. The contents of horizontal lists were explained in Chapter~14,
and the contents of vertical lists were explained in Chapter~15; it's time
now to describe what math lists are made of. Each item in a math list
is one of the following types of things:\enddanger

\smallskip
\item\bull an ^{atom} (to be explained momentarily);

\item\bull horizontal material (a rule or discretionary or penalty or
``whatsit'');

\item\bull vertical material (from |\mark| or |\insert| or |\vadjust|);

\item\bull a glob of ^{glue} (from |\hskip| or |\mskip| or |\nonscript|);

\item\bull a ^{kern} (from |\kern| or |\mkern|);

\item\bull a ^{style change} (from |\displaystyle|, |\textstyle|, etc.);

\item\bull a ^{generalized fraction} (from |\above|, |\over|, etc.);

\item\bull a ^{boundary} (from |\left| or |\right|);

\item\bull a four-way ^{choice} (from ^|\mathchoice|).

\ddanger The most important items are called {\sl atoms}, and they have
three parts: a {\sl^{nucleus}}, a {\sl^{superscript}}, and a {\sl^{subscript}}.
For example, if you type
\begintt
(x_i+y)^{\overline{n+1}}
\endtt
in math mode, you get a math list consisting of five atoms:
$($, $x_i$, $+$, $y$, and~$)^{\overline{n+1}}$. The nuclei of these atoms
are $($, $x$, $+$, $y$, and~$)$; the subscripts are empty except for the
second atom, which has subscript~$i$; the superscripts are empty except for the
last atom, whose superscript is~$\overline{n+1}$. This superscript is
itself a math list consisting of one atom, whose nucleus is~$n+1$; and that
nucleus is a math list consisting of three atoms.

\ddanger There are thirteen kinds of atoms, each of which might act
differently in a formula; for example, `$($' is an Open atom because
^^{atomic types, table}
it comes from an opening. Here is a complete list of the different kinds:
$$\halign{\indent#\hfil&\enskip#\hfil\cr
Ord&is an ordinary atom like `$x$'\thinspace;\cr
Op&is a large operator atom like `$\sum$'\thinspace;\cr
Bin&is a binary operation atom like `$+$'\thinspace;\cr
Rel&is a relation atom like `$=$'\thinspace;\cr
Open&is an opening atom like `$($'\thinspace;\cr
Close&is a closing atom like `$)$'\thinspace;\cr
Punct&is a punctuation atom like `$,$'\thinspace;\cr
Inner&is an inner atom produced by `|\left|$\,\ldots\,$|\right|';\cr
Over&is an overline atom like `$\overline x$'\thinspace;\cr
Under&is an underline atom like `$\underline x$'\thinspace;\cr
Acc&is an accented atom like `$\hat x$'\thinspace;\cr
Rad&is a radical atom like `$\sqrt2$'\thinspace;\cr
Vcent&is a vbox to be centered, produced by |\vcenter|.\cr
}$$

\ddanger An atom's nucleus, superscript, and subscript are called its
{\sl ^{fields}}, and there are four possibilities for each of these fields.
A field can be\enddanger

\smallskip
\item\bull empty;

\item\bull a math symbol (specified by family and position number);

\item\bull a box; or

\item\bull a math list.

\smallskip\noindent
For example, the Close atom $)^{\overline{n+1}}$ considered above has an
empty subscript field; its nucleus is the symbol `$)$', which is
character~\hex{28} of family~0 if the conventions of plain \TeX\ are
in force; and its superscript field is the math list $\overline{n+1}$.
The latter math list consists of an Over atom whose nucleus
is the math list $n+1$; and that math list, in turn, consists of
three atoms of types Ord, Bin, Ord.

\ddanger You can see \TeX's view of a math list by typing ^|\showlists|
in math mode. ^^{internal list format}
For example, after `|$(x_i+y)^{\overline{n+1}}\showlists|' your log
file gets the following curious data:
\begindisplay
|\mathopen|\cr
|.\fam0 (|\cr
|\mathord|\cr
|.\fam1 x|\cr
|_\fam1 i|\cr
\noalign{\penalty-500}
|\mathbin|\cr
|.\fam0 +|\cr
\noalign{\penalty-500}
|\mathord|\cr
|.\fam1 y|\cr
\noalign{\penalty-500}
|\mathclose|\cr
|.\fam0 )|\cr
|^\overline|\cr
|^.\mathord|\cr
|^..\fam1 n|\cr
|^.\mathbin|\cr
|^..\fam0 +|\cr
|^.\mathord|\cr
|^..\fam0 1|\cr
\enddisplay
In our previous experiences with |\showlists| we observed that there can
be boxes within boxes, and that each line in the log file is
prefixed by dots to indicate its position in the hierarchy. Math lists
have a slightly more complex structure; therefore a dot is used to denote
the nucleus of an atom, a~`|^|' is used for the superscript field, and
a~`|_|' is used for the subscript field. Empty fields are not shown. Thus,
for example, the Ord atom~$x_i$ is represented here by three lines
`|\mathord|', `|.\fam1 x|', and `|_\fam1 i|'.

\ddanger Certain kinds of atoms carry additional information besides their
nucleus, subscript, and superscript fields: An Op atom will be marked
`^|\limits|' or `^|\nolimits|' if the normal ^|\displaylimits|
convention has been overridden; a Rad atom contains
a delimiter field to specify what radical sign is to be used; and an Acc atom
contains the family and character codes of the accent symbol.

\ddanger When you say ^|\hbox||{...}| in math mode, an Ord atom is placed
on the current math list, with the hbox as its nucleus.  Similarly,
^|\vcenter||{...}| produces a Vcent atom whose nucleus is a box. But in
most cases the nucleus of an atom will be either a symbol or a math list.
You can experiment with |\showlists| to discover how other things like
fractions and mathchoices are represented internally.

\ddanger Chapter~26 contains complete details of how math lists are
constructed.  As soon as math mode ends (i.e., when the closing `|$|'
occurs), \TeX\ dismantles the current math list and converts it into a
horizontal list.  The rules for this conversion are spelled out in
Appendix~G\null. You can see ``before and after'' representations of such math
typesetting by ending a formula with `|\showlists$\showlists|'; the first
|\showlists| will display the math list, and the second will show the
(possibly complex) horizontal list that is manufactured from it.

\endchapter

The learning time is short. A few minutes gives the general flavor, and
typing a page or two of a paper generally uncovers most of the misconceptions.
\author ^{KERNIGHAN} and ^{CHERRY}, {\sl A System for %
  Typesetting Mathematics\/} (1975)
  % in Communications of the ACM p152

\bigskip

Within a few hours (a few days at most)
a typist with no math or typesetting experience
can be taught to input even the most complex equations.
\author PETER J. ^{BOEHM}, {\sl Software and Hardware Considerations %
  for a\break Technical Typesetting System\/} (1976)
%  in IEEE Transactions on Professional Communication PC-19, pp15--19

\eject
\beginchapter Chapter 18. Fine Points of\\Mathematics\\Typing

We have discussed most of the facilities needed to construct math
formulas, but there are several more things a good mathematical typist
will want to watch for. After you have typed a dozen or so formulas using
the basic ideas of Chapters 16 and~17, you will find that it's easy to
visualize the final appearance of a mathematical expression as you type
it.  And once you have gotten to that level, there's only a little bit
more to learn before you are producing formulas as beautiful as any the
world has ever seen; tastefully applied touches of \TeX nique will add a
professional polish that works wonders for the appearance and readability
of the books and papers that you type. This chapter talks about such
tricks, and it also fills in a few gaps by mentioning some aspects of math
that didn't fit comfortably into~Chapters~16~and~17.

\subsection Punctuation.  When a formula is followed by a ^{period}, ^{comma},
^{semicolon}, ^{colon}, ^{question mark}, ^{exclamation point}, etc., put the
^{punctuation} {\sl after\/} the |$|, when the formula is in the text; but
put the punctuation {\sl before\/} the |$$| when the formula is displayed.
For example,
\begintt
If $x<0$, we have shown that $$y=f(x).$$
\endtt
\TeX's spacing rules within paragraphs work best when the
^{punctuation marks} are not considered to be part of the formulas.

Similarly, don't ever type anything like
\begintt
for $x = a, b$, or $c$.
\endtt
It should be
\begintt
for $x = a$, $b$, or $c$.
\endtt
(Better yet, use a ^{tie}: `|or~$c$|'.) \ The reason is that \TeX\ will
typeset expression `|$x|~|=|~|a,|~|b$|' as a single formula, so it will
put a ``^{thin space}'' between the comma and the $b$. This space will
not be the same as the space that \TeX\ puts after the
comma {\sl after\/} the $b$, since spaces between words are always bigger than
thin spaces. Such unequal spacing looks bad, but when you type things right
the spacing will look good.

Another reason for not typing `|$x| |=| |a,| |b$|' is that it inhibits the
possibilities for breaking lines in a paragraph: \TeX\ will never break at
the space between the comma and the |b| because breaks after commas in
formulas are usually wrong. For example, in the equation
`|$x|~|=|~|f(a,|~|b)$|' we certainly don't want to put `$x=f(a,$' on one
line and `$b)$' on the next.

Thus, when typing formulas in the text of a paragraph, keep the math properly
segregated: Don't take operators like $-$ and $=$ outside of the |$|'s,
and keep commas inside the formula if they are truly part of the formula.
But if a comma or period or other punctuation mark belongs linguistically
to the sentence rather than to the formula, leave it outside the |$|'s.

\exercise Type this: $R(n,t)=O(t^{n/2})$, as $t\to0^+$.
\answer |$R(n,t)=O(t^{n/2})$, as $t\to0^+$.| \
(N.B.: `|O(|', not `|0('|.)

\danger Some mathematical styles insert a bit of extra space around
formulas to separate them from the text. For example, when copy is
being produced on an ordinary typewriter that doesn't have italic
letters, the best technical typists have traditionally put an extra
blank space before and after each formula, because this provides a
useful visual distinction. You might find it helpful to think of
each |$| as a symbol that has the potential of adding a little space
to the printed output; then the rule about excluding sentence
punctuation from formulas may be easier to remember.

\ddanger \TeX\ does, in fact, insert additional ^{space} before and after each
formula; the amount of such space is called ^|\mathsurround|, which is
a \<dimen>-valued parameter. For example, if you set |\mathsurround=1pt|,
each formula will effectively be 2~points wider ($1\pt$ at each side):
$$\baselineskip13pt\halign{\indent\mathsurround=#pt
  For $x=a$, $b$, or $c$.\hfil&\quad(#)\hfil\cr
1&|\mathsurround=1pt|\cr 0&|\mathsurround=0pt|\cr}$$
This extra space will disappear into the left or right margin if the formula
occurs at the beginning or end of a line. The value of\/ |\mathsurround| that
is in force when \TeX\ reads the closing |$| of a formula is used at both
left and right of that formula.  Plain \TeX\ takes |\mathsurround=0pt|, so
you won't see any extra space unless you are using some other format, or
unless you change |\mathsurround| yourself.

\subsection Non-italic letters in formulas. The names of algebraic variables
are usually italic or Greek letters, but common mathematical functions
like `log' are always set in ^{roman type}. The best way to deal with such
constructions is to make use of the following 32~control sequences (all of
which are defined in plain \TeX\ format, see Appendix~B):
\begintt
\arccos  \cos   \csc   \exp   \ker     \limsup  \min  \sinh
\arcsin  \cosh  \deg   \gcd   \lg      \ln      \Pr   \sup
\arctan  \cot   \det   \hom   \lim     \log     \sec  \tan
\arg     \coth  \dim   \inf   \liminf  \max     \sin  \tanh
\endtt
^^|\arccos| ^^|\cos| ^^|\csc| ^^|\exp| ^^|\ker| ^^|\limsup| ^^|\min|
^^|\sinh| ^^|\arcsin| ^^|\cosh| ^^|\deg| ^^|\gcd| ^^|\lg| ^^|\ln| ^^|\Pr|
^^|\sup| ^^|\arctan| ^^|\cot| ^^|\det| ^^|\hom| ^^|\lim| ^^|\log| ^^|\sec|
^^|\tan| ^^|\arg| ^^|\coth| ^^|\dim| ^^|\inf| ^^|\liminf| ^^|\max|
^^|\sin| ^^|\tanh|
These control sequences lead to roman type with appropriate
spacing:\def\sep{&\hskip-1em}
\beginlongmathdemo
\it Input\sep\it Output\cr
\noalign{\vskip2pt}
|$\sin2\theta=2\sin\theta\cos\theta$|\sep\sin2\theta=2\sin\theta\cos\theta\cr
|$O(n\log n\log\log n)$|\sep O(n\log n\log\log n)\cr
|$\Pr(X>x)=\exp(-x/\mu)$|\sep\Pr(X>x)=\exp(-x/\mu)\cr
|$$\max_{1\le n\le m}\log_2P_n$$|\sep
  \displaystyle{\max_{1\le n\le m}\log_2P_n}\cr
\noalign{\vskip2pt}
|$$\lim_{x\to0}{\sin x\over x}=1$$|\sep\displaystyle{\lim_{x\to0}
  {\sin x\over x}=1}\cr
\endmathdemo
^^|\mu|
The last two formulas, which are displays, show that some of the special
control sequences are treated by \TeX\ as ``large operators'' with limits
just like $\sum$: The subscript on |\max| is not treated like the subscript
on |\log|. Subscripts and superscripts will become limits
when they are attached to |\det|, |\gcd|, |\inf|, |\lim|, |\liminf|,
|\limsup|, |\max|, |\min|, |\Pr|, and |\sup|, in display style.

\exercise Express the following display in plain \TeX\ language, using
`^|\nu|' for `$\nu$':
$$p_1(n)=\lim_{m\to\infty}\sum_{\nu=0}^\infty
  \bigl(1-\cos^{2m}(\nu!^n\pi/n)\bigr).$$
\answer |$$p_1(n)=\lim_{m\to\infty}\sum_{\nu=0}^\infty|\parbreak
|  \bigl(1-\cos^{2m}(\nu!^n\pi/n)\bigr).$$|\par
\smallskip\noindent $\bigl[$Mathematicians may enjoy
interpreting this formula; cf.~G.~H. ^{Hardy},
{\sl Messenger of Mathematics\/ \bf35} (1906), 145--146.$\bigr]$

\danger If you need roman type for some mathematical function or operator
that isn't included in plain \TeX's list of~32, it is easy to define a
new control sequence by mimicking the definitions in Appendix~B\null. Or, if
you need roman type just for a ``one shot'' use, it is even easier to
get what you want by switching to ^|\rm| type, as follows:
\beginlongmathdemo
|$\sqrt{{\rm Var}(X)}$|&\sqrt{{\rm Var}(X)}\cr
|$x_{\rm max}-x_{\rm min}$|&x_{\rm max}-x_{\rm min}\cr
|${\rm LL}(k)\Rightarrow{\rm LR}(k)$|&{\rm LL}(k)\Rightarrow{\rm LR}(k)\cr
|$\exp(x+{\rm constant})$|&\exp(x+{\rm constant})\cr
|$x^3+{\rm lower\ order\ terms}$|&x^3+{\rm lower\ order\ terms}\cr
\endmathdemo
Notice the uses of `|\|\]' ^^{control space} in the last case;
without them, the result would have been `$x^3+{\rm lower order terms}$',
because ordinary blank ^{spaces} are ignored in math mode.

\danger You can also use ^|\hbox| instead of\/ |\rm| to get roman letters
into formulas. For example, four of the last five formulas can be
generated by
\beginlongmathdemo
|$\sqrt{\hbox{Var}(X)}$|&\sqrt{\hbox{Var}(X)}\cr
|$\hbox{LL}(k)\Rightarrow\hbox{LR}(k)$|&\hbox{LL}(k)\Rightarrow\hbox{LR}(k)\cr
|$\exp(x+\hbox{constant})$|&\exp(x+\hbox{constant})\cr
|$x^3+\hbox{lower order terms}$|&x^3+\hbox{lower order terms}\cr
\endmathdemo
In this case `|\|\]' isn't necessary, because the material in an
|\hbox| is processed in horizontal mode, when spaces are significant.
But such uses of\/ |\hbox| have two disadvantages: \ (1)~The contents of the
box will be typeset in the same size, whether or not the box occurs as a
subscript; for example, `|$x_{\hbox{max}}$|' yields `$x_{\hbox{max}}$'.
\ (2)~The font that's used inside |\hbox| will be the ``^{current font},''
so it might not be roman. For example, if you are typesetting the statement
of some theorem that is in slanted type, and if that theorem refers
to `|$\sqrt{\hbox{Var}(X)}$|', you will get the unintended result
`{\sl$\sqrt{\hbox{Var}(X)}$}'. In order to make sure that an |\hbox| uses
roman type, you need to specify |\rm|, e.g., `|$\sqrt{\hbox{\rm Var}(X)}$|';
and then the |\hbox| serves no purpose. We will see later, however, that
|\hbox| can be very useful in displayed formulas.

\ddangerexercise When the displayed formula
`|$$\lim_{n\to\infty}x_n {\rm\ exists} \iff|\break
|\limsup_{n\to\infty}x_n = \liminf_{n\to\infty}x_n.$$|' is typeset with
the standard macros of plain \TeX, you get
$$\lim_{n\to\infty}x_n{\rm\ exists}\iff
  \limsup_{n\to\infty}x_n=\liminf_{n\to\infty}x_n.$$
But some people prefer a different notation: Explain how you could change
the definitions of\/ ^|\limsup| and ^|\liminf| so that the display would be
$$
\def\limsup{\mathop{\overline{\rm lim}}}
\def\liminf{\mathop{\underline{\rm lim}}}
\lim_{n\to\infty}x_n{\rm\ exists}\iff
  \limsup_{n\to\infty}x_n=\liminf_{n\to\infty}x_n.$$
\answer |\def\limsup{\mathop{\overline{\rm lim}}}|\parbreak
|\def\liminf{\mathop{\underline{\rm lim}}}|\par
\smallskip\noindent
[Notice that the limits `$n\to\infty$' appear at different levels, in both
of the displays, because `sup' and the underbar descend below the baseline.
It is possible to unify the limit positions by using ^{phantoms}, as explained
later in this chapter. For example,
\begintt
\def\limsup{\mathop{\vphantom{\underline{}}\overline{\rm lim}}}
\endtt
would give lower limits in the same position as |\liminf|.]

\danger The word `mod' is also generally set in roman type, when it occurs
in formulas; but this word needs more care, because it is used in two
different ways that require two different treatments.
Plain \TeX\ provides two different control sequences,
^|\bmod| and ^|\pmod|, for the two cases: |\bmod| is to be used when
`mod' is a ^{binary operation} (i.e., when it occurs between two quantities,
like a plus sign usually does), and |\pmod| is to be used when
`mod' occurs parenthetically at the end of a formula. For example,
\beginmathdemo
|$\gcd(m,n)=\gcd(n,m\bmod n)$|&\gcd(m,n)=\gcd(n,m\bmod n)\cr
|$x\equiv y+1\pmod{m^2}$|&x\equiv y+1\pmod{m^2}\cr
\endmathdemo
The `|b|' in `|\bmod|' stands for ``binary''; the `|p|' in `|\pmod|' stands
for ``parenthesized.'' Notice that |\pmod| inserts its own parentheses;
the quantity that appears after `mod' in the parentheses should be
enclosed in braces, if it isn't a single symbol.

\dangerexercise What did poor B. L. ^{User} get when he typed
`|$x\equiv0 (\pmod y^n)$|'\thinspace?
\answer $x\equiv0(\pmod y^n)$. He should have typed
`|$x\equiv0\pmod{y^n}$|'.

\dangerexercise Explain how to produce \lower12pt\null\
$\smash{\displaystyle{n\choose k}\equiv{\lfloor n/p\rfloor\choose
\lfloor k/p\rfloor}{n\bmod p\choose k\bmod p}\pmod p.}$
\answer |$${n\choose k}\equiv{\lfloor n/p\rfloor\choose|\parbreak
|  \lfloor k/p\rfloor}{n\bmod p\choose k\bmod p}\pmod p.$$|

\danger The same mechanism that works for roman type in formulas can be used
to get other styles of type as well. For example, ^|\bf| yields ^{boldface}:
\beginmathdemo
|$\bf a+b=\Phi_m$|&\bf a+b=\Phi_m\cr
\endmathdemo
Notice that whole formula didn't become emboldened in this example; the
`$+$' and `$=$' stayed the same. Plain \TeX\ sets things up so
that commands like |\rm| and |\bf| will affect only the uppercase letters
|A|~to~|Z|, the lowercase letters |a|~to~|z|, the digits |0|~to~|9|,
the uppercase Greek letters |\Gamma| to~|\Omega|, and math ^{accents}
like ^|\hat| and ^|\tilde|.  Incidentally, no braces were used in this
example, because |$|'s have the effect of grouping; |\bf| changes the
current font, but the change is local, so it does not affect the font that
was current outside the formula.

\ddanger The bold fonts available in plain \TeX\ are ``bold roman,'' rather
than ``bold italic,'' because the latter are rarely needed. However, \TeX\
could readily be set up to make use of bold math italics, if desired
(see exercise 17.\bmiexno). A more extensive set of math fonts would also
include ^{script}, ^{Fraktur}, and ``^{blackboard bold}'' styles; plain
\TeX\ doesn't have these, but other formats like \AmSTeX\ do. ^^{AMS-TeX}
^^{German black letters}

\danger Besides |\rm| and |\bf|, you can say ^|\cal| in formulas to get
uppercase letters in a ``^{calligraphic}'' style. For example, `|$\cal
A$|' produces `$\cal A$' and `|$\cal Z$|' produces `$\cal Z$'. But beware:
This works only with the letters |A| to |Z|; you'll get weird results if
you apply |\cal| to lowercase or Greek letters.

\danger There's also ^|\mit|, which stands for ``^{math italic}.'' This
affects ^{uppercase Greek}, so that you get
$\mit(\Gamma,\Delta,\Theta,\Lambda,\Xi,\Pi,\Sigma,\Upsilon,\Phi,\Psi,\Omega)$
instead of $(\Gamma,\ldots,\Omega)$.  When~|\mit| is in effect, the
ordinary letters |A| to |Z| and |a| to |z| are not changed; they are set
in italics as usual, because they ordinarily come from the math italic
font. Conversely, uppercase Greek letters and math accents are unaffected
by |\rm|, because they ordinarily come from the roman font. Math accents
should not be used when the |\mit| family has been selected, because the
math italic font contains no accents.

\dangerexercise Type the formula $\bf\bar x^{\rm T}Mx={\rm0}\iff x=0$,
using as few keystrokes as possible.  ^^{boldface numbers in math}
\ (The first `0' is roman, the second is bold. The superscript `T' is roman.)
\answer |$\bf\bar x^{\rm T}Mx={\rm0}\iff x=0$|.  \ (If you typed a space between
|\rm| and~|0|, you wasted a keystroke; but don't feel guilty about it.)

\dangerexercise Figure out how to typeset
`$S\subseteq\mit\Sigma\iff S\in\cal S$'.
\answer |$S\subseteq{\mit\Sigma}\iff S\in{\cal S}$|. In this case the
braces are redundant and could be eliminated; but you shouldn't try to do
{\sl everything\/} with fewest keystrokes, or you'll outsmart yourself
some day.

\danger Plain \TeX\ also allows you to type ^|\it|, ^|\sl|, or ^|\tt|, if
you want text italic, slanted, or typewriter letters to occur in a math
formula. However, these fonts are available only in text size, so you
should not try to use them in subscripts.

\danger If you're paying attention, you probably wonder why both
|\mit| and |\it| are provided; the answer is that |\mit| is ``math italic''
(which is normally best for formulas), and |\it| is ``^{text italic}'' (which
is normally best for running text).
\beginmathdemo
|$This\ is\ math\ italic.$|&This\ is\ math\ italic.\cr
|{\it This is text italic.}|&\hbox{\it This is text italic.}\cr
\endmathdemo
The math italic letters are a little wider, and the spacing is different;
this works better in most formulas, but it fails spectacularly when
you try to type certain italic words like `$different$' using math mode
(`|$different$|'). A wide `$f$' is usually desirable in formulas, but it
is undesirable in text. Therefore wise typists
use |\it| in a math formula that is supposed
to contain an actual italic word. Such cases almost never occur in
classical mathematics, but they are common when ^{computer programs}
are being typeset, since programmers often use multi-letter ``^{identifiers}'':
\beginmathdemo
|$\it last:=first$|&\it last:=first\cr
|$\it x\_coord(point\_2)$|&\it x\_coord(point\_2)\cr
\endmathdemo
The first of these examples shows that \TeX\ recognizes the ^{ligature}
`{\it fi\/}' when text italic occurs in a math formula;
the other example illustrates the use of short ^{underlines} to break
up identifier names. ^^{control-underline}
When the author typeset this manual, he used `|$\it SS$|' to refer to
style~$\SS$, since `|$SS$|' makes the $S$'s too far apart: $SS$.

\dangerexercise What plain \TeX\ commands will produce the following display?
$$\tenmath
{\it available}+\sum_{i=1}^n\max\bigl({\it full}(i),{\it reserved}(i)\bigr)
  ={\it capacity}.$$
\answer |$${\it available}+\sum_{i=1}^n\max\bigl({\it full}(i),|\parbreak
        |{\it reserved}(i)\bigr)={\it capacity}.$$|
\smallskip\noindent [If\/ |\it| had been used throughout
the formula, the subscript~$i$ and superscript~$n$ would have caused error
messages saying `^|\scriptfont| |4| |is| |undefined|',
since plain \TeX\ makes |\it| available only in text size.]

\ddangerexercise How would you go about typesetting the following computer
program, using the macros of plain \TeX?
$$\vbox{\let\par=\endgraf
\obeylines\sfcode`;=3000
{\bf for $j:=2$ step $1$ until $n$ do}
\quad {\bf begin} ${\it accum}:=A[j]$; $k:=j-1$; $A[0]:=\it accum$;
\quad {\bf while $A[k]>\it accum$ do}
\qquad {\bf begin} $A[k+1]:=A[k]$; $k:=k-1$;
\qquad {\bf end};
\quad $A[k+1]:=\it accum$;
\quad {\bf end}.
}$$
\answer |{\obeylines \sfcode`;=3000|^^|\sfcode|\parbreak
|{\bf for $j:=2$ step $1$ until $n$ do}|\parbreak
|\quad {\bf begin} ${\it accum}:=A[j]$; $k:=j-1$; $A[0]:=\it accum$;|\parbreak
|\quad {\bf while $A[k]>\it accum$ do}|\parbreak
|\qquad {\bf begin} $A[k+1]:=A[k]$; $k:=k-1$;|\parbreak
|\qquad {\bf end};|\parbreak
|\quad $A[k+1]:=\it accum$;|\parbreak
|\quad {\bf end}.\par}|\par
\smallskip\noindent
[This is something like the ``poetry'' example in Chapter~14, but much
more difficult. Some manuals of style say that ^{punctuation} should inherit
the font of the preceding character, so that three kinds of semicolons
should be typeset; e.g., these experts recommend `$k:=j-1$; \
$A[0]:={}${\it accum;} \ {\bf end;}'. The author heartily disagrees.]

\subsection Spacing between formulas. ^{Displays} often contain more than one
formula; for example, an equation is frequently accompanied by a ^{side
condition}:
$$F_n=F_{n-1}+F_{n-2},\qquad n\ge2.$$
In such cases you need to tell \TeX\ how much space to put after the comma,
because \TeX's normal spacing conventions would bunch things together;
without special precautions you would get
$$F_n=F_{n-1}+F_{n-2}, n\ge2.$$

The traditional hot-metal technology for printing has led to some ingrained
standards for situations like this, based on what printers call a ``^{quad}''
of space. Since these standards seem to work well in practice, \TeX\ makes
it easy for you to continue the tradition: When you type `^|\quad|' in plain
\TeX\ format, you get a printer's quad of space in the horizontal direction.
Similarly, `^|\qquad|' gives you a double quad (twice as much); this
is the normal spacing for situations like
the $F_n$ example above. Thus, the recommended procedure is to type
\begintt
$$ F_n = F_{n-1} + F_{n-2}, \qquad  n \ge 2. $$
\endtt
It is perhaps worth reiterating that \TeX\ ignores all the spaces in math
mode (except, of course, the space after `|\qquad|', which is needed
to distinguish between `|\qquad|~|n|' and `|\qquadn|'); so the same result
would be obtained if you were to leave out all but one space:
\begintt
$$F_n=F_{n-1}+F_{n-2},\qquad n\ge2.$$
\endtt
Whenever you want spacing that differs from the normal conventions, you must
specify it explicitly by using control sequences such as |\quad| and |\qquad|.

\danger A quad used to be a square piece of blank type, $1\em$ wide and $1\em$
tall---approximately the size of a capital M, as explained in Chapter~10. This
tradition has not been fully retained: The control sequence |\quad| in plain
\TeX\ is simply an abbreviation for `|\hskip|~|1|^|em||\relax|', so \TeX's
quad has width but no height.

\danger You can use |\quad| in text as well as in formulas; for example,
Chapter~14 illustrates how |\quad| applies to poetry. When |\quad| appears
in a formula it stands for one~em in the current text font, independent of the
current math size or style or family. Thus, for example, |\quad| is just
as wide in a subscript as it is on the main line of a formula.

Sometimes a careless author will put two formulas next to each other in
the text of a paragraph. For example, you might find a sentence like this:
\begindisplay
The ^{Fibonacci} numbers satisfy $F_n=F_{n-1}+F_{n-2}$, \ $n\ge2$.
\enddisplay
Everybody who teaches proper ^{mathematical} ^{style} is agreed that formulas
ought to be separated by words, not just by commas; the author of that
sentence should at least have said `for $n\ge2$', not simply `$n\ge2$'.
But alas, such lapses are commonplace, and many prominent mathematicians
are hopelessly addicted to clusters of formulas. If we are not allowed to change
their writing style, we can at least insert extra space where they
neglected to insert an appropriate word. An additional interword space
generally works well in such cases; for example, the sentence above was
typeset thus:
\begintt
... $F_n=F_{n-1}+F_{n-2}$, \ $n\ge2$.}$$
\endtt
The `|\|\]' ^^{control space} here gives a visual separation that
partly compensates for the bad style.

\exercise Put the following paragraph into \TeX\ form, treating punctuation
and spacing carefully; also insert ^{ties} to prevent bad line breaks.
\begindisplay\baselineskip13pt
\vbox{\raggedright\hsize=310pt\parindent=0pt
Let $H$~be a Hilbert space, \
$C$~a closed bounded convex subset of~$H$, \
$T$~a nonexpansive self map of~$C$.
Suppose that as $n\to\infty$, \ $a_{n,k}\to0$ for each~$k$,
and $\gamma_n=\sum_{k=0}^\infty(a_{n,k+1}-a_{n,k})^+\to0$.
Then for each $x$~in~$C$, \
$A_nx=\sum_{k=0}^\infty a_{n,k}T^kx$ converges weakly
to a fixed point of~$T$.
} % taken from Bull. AMS 82 (1976), p 959; chosen by AMS in '78 for demo
\enddisplay
\answer |Let $H$~be a Hilbert space, \
$C$~a closed bounded convex subset of~$H$, \
$T$~a nonexpansive self map of~$C$.
Suppose that as $n\to\infty$, \ $a_{n,k}\to0$ for each~$k$,
and $\gamma_n=\sum_{k=0}^\infty(a_{n,k+1}-|\allowbreak|a_{n,k})^+\to0$.
 Then for each $x$~in~$C$,  \
$A_nx=\sum_{k=0}^\infty a_{n,k}T^kx$ converges weakly
to a fixed point of~$T$.|\par
[If any mathematicians are reading this, they might either appreciate
or resent the following attempt to edit the given paragraph
into a more acceptable style: ``%
Let $C$~be a closed, bounded, convex subset of a Hilbert space~$H$,
and let $T$~be a nonexpansive self map of~$C$.
Suppose that as $n\to\infty$, we have $a_{n,k}\to0$ for each~$k$,
and $\gamma_n=\sum_{k=0}^\infty(a_{n,k+1}-a_{n,k})^+\to0$.
Then for each $x$~in~$C$, the infinite sum
$A_nx=\sum_{k=0}^\infty a_{n,k}T^kx$ converges weakly
to a fixed point of~$T$.'']

\subsection Spacing within formulas. Chapter 16 says that \TeX\ does
automatic ^{spacing} of math formulas so that they look right, and this is
almost true. But occasionally you must give \TeX\ some help. The number of
possible math formulas is vast, and \TeX's spacing rules are rather
simple, so it is natural that exceptions should arise.  Of course, it is
desirable to have fine units of spacing for this purpose, instead of the
big chunks that arise from |\|\], |\quad|, and |\qquad|.

The basic elements of space that \TeX\
puts into formulas are called {\sl ^{thin spaces}}, {\sl ^{medium
spaces}}, and {\sl ^{thick spaces}}.  In order to get a feeling for these
units, let's take a look at the $F_n$ example again: Thick spaces occur
just before and after the = sign, and also before and after the $\ge$\thinspace;
medium spaces occur just before and after the $+$ sign.  Thin spaces are
slightly smaller, but noticeable; it's a thin space that makes the
difference between `loglog' and `$\log\log$'. The normal space between
words of a paragraph is approximately equal to two thin spaces.

\TeX\ inserts thin spaces, medium spaces, and thick spaces into formulas
automatically, but you can add your own spacing whenever you want to,
by using the control sequences ^^|\,|^^|\!|^^|\;|^^|\>|
$$\halign{\indent#\hfil&\quad#\hfil\cr
|\,|&thin space \ (normally 1/6 of a quad);\cr
|\>|&medium space \ (normally 2/9 of a quad);\cr
|\;|&thick space \ (normally 5/18 of a quad);\cr
|\!|&negative thin space \ (normally $-1/6$ of a quad).\cr}$$
In most cases you can rely on \TeX's spacing while you are typing a manuscript,
and you'll want to insert or delete space with these four control sequences
only in rare circumstances after you see what comes out.

\ddanger We observed a minute ago that |\quad| spacing does not
change with the style of formula, nor does it depend on the math font
families that are being used. But thin spaces, medium spaces, and thick
spaces do get bigger and smaller as the size of type gets bigger and
smaller; this is because they are defined in terms of ^\<muglue>, a~special
brand of glue intended for math spacing. You specify \<muglue> just
as if it were ordinary glue, except that the units are given in terms of
`^|mu|' (math units) instead of~|pt| or~|cm| or something else. For
example, Appendix~B contains the definitions
\begintt
\thinmuskip = 3mu
\medmuskip = 4mu plus 2mu minus 4mu
\thickmuskip = 5mu plus 5mu
\endtt
^^|\thinmuskip|^^|\medmuskip|^^|\thickmuskip|
and this defines the thin, medium, and thick spaces that \TeX\ inserts
into formulas. According to these specifications, thin spaces in plain
\TeX\ do not stretch or shrink; medium spaces can stretch a little, and
they can shrink to zero; thick spaces can stretch a lot, but they never shrink.

\ddanger There are 18 mu to an em, where the em is taken from family~2
(the math symbols family). In other words, ^|\textfont|~|2| defines the em
value for |mu| in display and text styles; ^|\scriptfont|~|2| defines the
em for script size material; and ^|\scriptscriptfont|~|2| defines it for
scriptscript size.

\ddanger You can insert math glue into any formula just by giving
the command `^|\mskip|\<muglue>'. For example, `|\mskip 9mu plus 2mu|'
inserts one half em of space, in the current size, together with some
stretchability. Appendix~B defines `|\,|' to be an abbreviation for
`|\mskip\thinmuskip|'.  Similarly, you can use the command `^|\mkern|'
when there is no stretching or shrinking; `|\mkern18mu|' gives one em of
horizontal space in the current size. \TeX\ insists that |\mskip| and
|\mkern| be used only with |mu|; conversely, ^|\hskip| and ^|\kern| (which
are also allowed in formulas) must never give units in |mu|.

Formulas involving ^{calculus} look best when an extra thin space appears
before $dx$ ^^{dx} or~$dy$ or~$d\,$whatever; but \TeX\ doesn't do this
automatically. Therefore a well-trained typist will remember to insert
`|\,|'  in examples like the following:
\beginmathdemo
\it Input&\it Output\cr
\noalign{\vskip2pt}
|$\int_0^\infty f(x)\,dx$|&\int_0^\infty f(x)\,dx\cr
|$y\,dx-x\,dy$|&y\,dx-x\,dy\cr
|$dx\,dy=r\,dr\,d\theta$|&dx\,dy=r\,dr\,d\theta\cr
|$x\,dy/dx$|&x\,dy/dx\cr \endmathdemo Notice that no `|\,|' was desirable
after the `|/|' in the last example.  Similarly, there's no need for
`|\,|' in cases like
\begindisplaymathdemo
|$$\int_1^x{dt\over t}$$|&\int_1^x{dt\over t}\cr
\endmathdemo
since the $dt$ appears all by itself in the numerator of a fraction; this
detaches it visually from the rest of the formula.

\exercise Explain how to handle the display
$$\int_0^\infty{t-ib\over t^2+b^2}e^{iat}\,dt=e^{ab}E_1(ab),\qquad a,b>0.$$
\answer |$$\int_0^\infty{t-ib\over t^2+b^2}e^{iat}\,dt=|\parbreak
        |    e^{ab}E_1(ab),\qquad a,b>0.$$|

\danger When physical ^{units} appear in a formula, they should be set in roman
type and separated from the preceding material by a thin space:
\beginmathdemo
|$55\rm\,mi/hr$|&55\rm\,mi/hr\cr
|$g=9.8\rm\,m/sec^2$|&g=9.8\rm\,m/sec^2\cr
|$\rm1\,ml=1.000028\,cc$|&\rm1\,ml=1.000028\,cc\cr
\endmathdemo

\dangerexercise Typeset the following display, assuming that `^|\hbar|'
generates `$\hbar$':
$$\hbar=1.0545\times10^{-27}\rm\,erg\,sec.$$
\answer |$$\hbar=1.0545\times10^{-27}\rm\,erg\,sec.$$|

\danger Thin spaces should also be inserted after ^{exclamation points}
(which stand for the ``^{factorial}'' operation in a formula), if the next
character is a letter or a number or an opening delimiter:
\beginmathdemo
|$(2n)!/\bigl(n!\,(n+1)!\bigr)$|&(2n)!/\bigl(n!\,(n+1)!\bigr)\cr
\noalign{\vskip6pt}
|$${52!\over13!\,13!\,26!}$$|&\displaystyle{52!\over13!\,13!\,26!}\cr
\endmathdemo

Besides these cases, you will occasionally encounter formulas in which
the symbols are bunched up too tightly, or where too much white space
appears, because of certain unlucky combinations of shapes. It's usually
impossible to anticipate optical glitches like this until you see the first
proofs of what you have typed; then you get to use your judgment about how
to add finishing touches that provide extra beauty, clarity, and finesse.
A tastefully applied `|\,|' or `|\!|'\ will open things up or close things
together so that the reader won't be distracted from the mathematical
significance of the formula. ^{Square root} signs and ^{multiple integrals} are
often candidates for such fine tuning. Here are some examples of situations
to look out for:
\beginmathdemo
|$\sqrt2\,x$|&\sqrt2\,x\cr
|$\sqrt{\,\log x}$|&\sqrt{\,\log x}\cr
|$O\bigl(1/\sqrt n\,\bigr)$|&O\bigl(1/\sqrt n\,\bigr)\cr
|$[\,0,1)$|&[\,0,1)\cr
|$\log n\,(\log\log n)^2$|&\log n\,(\log\log n)^2\cr
|$x^2\!/2$|&x^2\!/2\cr
|$n/\!\log n$|&n/\!\log n\cr
|$\Gamma_{\!2}+\Delta^{\!2}$|&\Gamma_{\!2}+\Delta^{\!2}\cr
|$R_i{}^j{}_{\!kl}$|&R_i{}^j{}_{\!kl}\cr
|$\int_0^x\!\int_0^y dF(u,v)$|&\int_0^x\!\int_0^y dF(u,v)\cr
\noalign{\vskip6pt}
|$$\int\!\!\!\int_D dx\,dy$$|&\displaystyle{\int\!\!\!\int_D dx\,dy}\cr
\endmathdemo
^^|\Gamma|^^|\Delta|^^|\intint|
In each of these formulas the omission of\/ |\,| or |\!|\ would lead to
somewhat less satisfactory results.

\ddanger Most of these examples where thin-space corrections are desirable
arise because of chance coincidences. For example, the superscript in
|$x^2/2$| leaves a hole before the slash ($x^2/2$); a negative thin
space helps to fill that hole. The positive thin space in |$\sqrt{\,\log x}$|
compensates for the fact that `$\log x$' begins with a tall, unslanted
letter; and so on. But two of the examples involve corrections that
were necessary because \TeX\ doesn't really know a great deal about
mathematics: \ (1)~In the formula
|$\log n(\log\log n)^2$|, \TeX\ inserts no thin space before the left
parenthesis, because there are similar formulas like |$\log n(x)$| where
no such space is desired. \ (2)~In the formula |$n/\log n$|, \TeX\
automatically inserts an unwanted thin~space before |\log|, since the slash is
treated as an ordinary symbol, and since a~thin space is usually desirable
between an ordinary symbol and an operator like |\log|.

\ddanger In fact, \TeX's rules for spacing in formulas are fairly simple.
A formula is converted to a math list as described at the end of Chapter~17,
and the math list consists chiefly of ``^{atoms}'' of eight basic types:
^{Ord}~(^{ordinary}), ^{Op}~(^{large operator}), ^{Bin}~(^{binary operation}),
^{Rel}~(^{relation}), ^{Open}~(^{opening}), ^{Close}~(^{closing}),
^{Punct}~(punctuation), ^^{punctuation} and ^{Inner}~(a delimited
subformula).  Other kinds of atoms, which arise from commands like
^|\overline| or ^|\mathaccent| or ^|\vcenter|, etc., are all treated as
type~Ord; subformulas delimited by ^|\left| and ^|\right|
are treated as type~Inner. The following table is
used to determine the spacing between pairs of adjacent atoms:
$$\baselineskip0pt\lineskip0pt
\halign to\hsize
 {\strut\hbox to\parindent{\it#\hfil}& % for the legend "Left atom"
  #\hfil\quad& % for the row labels
  #\hfil\tabskip 0pt plus 10pt& % for the rule at the left
  \hbox to 25pt{\tt\hss#\hss}& % for column 1
  \hbox to 25pt{\tt\hss#\hss}& % for column 2
  \hbox to 25pt{\tt\hss#\hss}& % for column 3
  \hbox to 25pt{\tt\hss#\hss}& % for column 4
  \hbox to 25pt{\tt\hss#\hss}& % for column 5
  \hbox to 25pt{\tt\hss#\hss}& % for column 6
  \hbox to 25pt{\tt\hss#\hss}& % for column 7
  \hbox to 25pt{\tt\hss#\hss}& % for column 8
  #\hfil\tabskip0pt\cr % for the rule at the right
&&&&\multispan7\hss\it Right atom\hss\cr
\noalign{\vskip3pt}
&&&\rm Ord&\rm Op&\rm Bin&\rm Rel&\rm Open&\rm Close&\rm Punct&\rm Inner\cr
\noalign{\vskip2pt}
\omit&&\multispan{10}\leaders\hrule\hfil\cr
\omit\vbox to 2pt{}&&\vrule&&&&&&&&&\vrule\cr
&Ord&\vrule&0&1&(2)&(3)&0&0&0&(1)&\vrule\cr
&Op&\vrule&1&1&*&(3)&0&0&0&(1)&\vrule\cr
&Bin&\vrule&(2)&(2)&*&*&(2)&*&*&(2)&\vrule\cr
Left&Rel&\vrule&(3)&(3)&*&0&(3)&0&0&(3)&\vrule\cr
atom&Open&\vrule&0&0&*&0&0&0&0&0&\vrule\cr
&Close&\vrule&0&1&(2)&(3)&0&0&0&(1)&\vrule\cr
&Punct&\vrule&(1)&(1)&*&(1)&(1)&(1)&(1)&(1)&\vrule\cr
&Inner&\vrule&(1)&1&(2)&(3)&(1)&0&(1)&(1)&\vrule\cr
\omit\vbox to 2pt{}&&\vrule&&&&&&&&&\vrule\cr
\omit&&\multispan{10}\leaders\hrule\hfil\cr}$$
^^{spacing table} ^^{math spacing table}
Here 0, 1, 2, and 3 stand for no space, thin space, medium space, and
thick space, respectively; the table entry is parenthesized if the space
is to be inserted only in display and text styles, not in script and
scriptscript styles. For example, many of the entries in the Rel row
and the Rel column are `{\tt(3)}'; this means that thick spaces are normally
inserted before and after relational symbols like `=', but not in
subscripts. Some of the entries in the table are `{\tt*}'; such cases
never arise, because Bin atoms must be preceded and followed by atoms
compatible with the nature of binary operations. Appendix~G contains
precise details about how math lists are converted to horizontal lists;
this conversion is done whenever \TeX\ is about to leave math mode, and the
interatomic spacing is inserted at that time.

\ddanger For example, the displayed formula specification
\begintt
$$x+y=\max\{x,y\}+\min\{x,y\}$$
\endtt
will be transformed into the sequence of atoms
\def\\#1{\vbox to 33pt{\vbox to 22pt{\vfill\hrule
      \hbox{\vrule\hskip-.4pt$#1$\hskip-.4pt\vrule}}\hrule\vfill}}%
\begindisplay
\vbox{\vskip-11pt\hbox{$
\\x\;\;\\+\;\;\\y\;\;\\=\;\;\\\max\;\;\\\{\;\;
\\x\;\;\\,\;\;\\y\;\;\\\}\;\;\\+\;\;\\\min\;\;\\\{
\;\;\\x\;\;\\,\;\;\\y\;\;\\\}$}\vskip-11pt}
\enddisplay
of respective types Ord, Bin, Ord, Rel, Op, Open, Ord, Punct, Ord, Close,
Bin, Op, Open, Ord, Punct, Ord, and Close.
Inserting spaces according to the table gives
$$\def\0{\thinspace}
\def\1{\thinspace{\tt\bslash,}\thinspace}
\def\2{\thinspace{\tt\bslash>}\thinspace}
\def\3{\thinspace{\tt\bslash;}\thinspace}
\halign{\indent\hfil#\cr
  Ord\2Bin\2Ord\3Rel\3Op\0Open\0Ord\0Punct\1Ord\0Close\2\qquad\cr
                 Bin\2Op\0Open\0Ord\0Punct\1Ord\0Close\cr}$$
and the resulting formula is
$$\vbox{\vskip-11pt\hbox{$
\\x\>\\+\>\\y\;\\=\;\\\max\\\{
\\x\\,\,\\y\\\}\>\\+\>\\\min\\\{
\\x\\,\,\\y\\\}$}\vskip-11pt}$$
i.e.,$$x+y=\max\{x,y\}+\min\{x,y\}\rlap{\quad.}$$
This example doesn't involve subscripts or superscripts; but subscripts and
superscripts merely get attached to atoms without changing the atomic type.

\ddangerexercise Use the table to determine what spacing \TeX\
will insert between the atoms of the formula `|$f(x,y)<x^2+y^2$|'.
\answer There are ten atoms (the first is $f$ and last is $y^2$); their types,
and the interatomic spacing, are respectively
\begindisplay \def\0{\thinspace}%
  \def\1{\thinspace{\tt\bslash,}\thinspace}%
  \def\2{\thinspace{\tt\bslash>}\thinspace}%
  \def\3{\thinspace{\tt\bslash;}\thinspace}
Ord\0Open\0Ord\0Punct\1Ord\0Close\3Rel\3Ord\2Bin\2Ord.
\enddisplay

\ddanger The plain \TeX\ macros ^|\bigl|, ^|\bigr|, ^|\bigm|, and ^|\big|
all produce identical delimiters; the only difference between them is that
they may lead to different spacing, because they make the delimiter into
different types of atoms: |\bigl| produces an Open atom, |\bigr| a~Close,
|\bigm| a~Rel, and |\big| an~Ord. On the other hand, when a subformula
appears between ^|\left| and ^|\right|, it is typeset by itself and placed
into an Inner atom. Therefore it is possible that a subformula enclosed by
|\left| and |\right| will be surrounded by more space than there would be
if that subformula were enclosed by |\bigl| and~|\bigr|. For example, Ord
followed by Inner (from |\left|) gets a thin space, but Ord followed by
Open (from |\bigl|) does not. The rules in Chapter~17 imply that the
construction `^|\mathinner||{\bigl({|\<subformula>|}\bigr)}|' within any
formula produces a result essentially equivalent to
`|\left(|\<subformula>|\right)|', when the \<subformula> doesn't end
with Punct, except that the ^{delimiters} are forced to
be of the |\big| size regardless of the height and depth of the subformula.

\danger \TeX's spacing rules sometimes fail when `\|' and `|\|\|'
appear in a formula, because $\vert$ and $\Vert$ are treated as ordinary symbols
^^{verticalline}^^|\verticalline|
instead of as delimiters. For example, consider the formulas
\beginlongmathdemo
|$|\||-x|\||=|\||+x|\||$|&\vert-x\vert=\vert+x\vert\cr
|$\left|\||-x\right|\||=\left|\||+x\right|\||$|&
  \left\vert-x\right\vert=\left\vert+x\right\vert\cr
|$\lfloor-x\rfloor=-\lceil+x\rceil$|&\lfloor-x\rfloor=-\lceil+x\rceil\cr
\endmathdemo
In the first case the spacing is wrong because \TeX\ thinks that the plus
sign is computing the sum of `$\vert$' and `$x$'. The use of\/ |\left| and
|\right| in the second example puts \TeX\ on the right track.  The third
example shows that no such corrections are needed with other delimiters,
because \TeX\ knows whether they are openings or closings.

\ddangerexercise Some perverse mathematicians use ^{brackets} backwards,
to denote ``^{open intervals}.''
Explain how to type the following bizarre formula: % from MR53 #3451
$\left]-\infty,T\right[\times\left]-\infty,T\right[$.
\answer |$\left]-\infty,T\right[\times\left]-\infty,T\right[$|. \ (Or one could
say ^|\mathopen| and ^|\mathclose| instead of\/ |\left| and |\right|;
then \TeX\ would not choose the size of the delimiters, nor would it consider
the subformulas to be of type Inner.) \
% that formula was quoted from MR review of paper by Mario Marino
% in Ricerche Mat. 24 (1975), no.~1, 152--171
Open intervals are more clearly expressed in print
by using parentheses instead of reversed brackets; for example,
compare `$(-\infty,T)\times(-\infty,T)$' to the given formula.

\ddangerexercise Study Appendix G and determine what spacing will be used
in the formula `|$x++1$|'. Which of the plus signs will be regarded
as a ^{binary operation}?
\answer The first |+| will become a Bin atom, the second an Ord; hence
the result is $x$, medium space, $+$, medium space, $+$, no space, 1.

\subsection Ellipses\/ {\rm(``three dots'')}. ^^{ellipses}
Mathematical copy looks much nicer if you are careful about how groups of
^{three dots} are typed in formulas and text. Although it looks fine to
type `|...|'\ on a typewriter that has fixed spacing, the result looks too
crowded when you're using a printer's fonts:  `|$x...y$|' results in
`$x...y$', and such close spacing is undesirable except in subscripts or
superscripts.

An ellipsis can be indicated by two different kinds of dots, one higher
than the other; the best mathematical traditions distinguish between
these two possibilities. It is generally correct to produce formulas like
\begindisplay
$\displaystyle x_1+\cdots+x_n\qquad {\rm and}\qquad (x_1,\ldots,x_n),$
\enddisplay
but wrong to produce formulas like
\begindisplay
$\displaystyle x_1+\ldots+x_n\qquad {\rm and}\qquad (x_1,\cdots,x_n).$
\enddisplay
The plain \TeX\ format of Appendix B allows you to solve the ``three dots''
problem very simply, and everyone will be envious of the beautiful formulas
that you produce. The idea is simply to type ^|\ldots| when you want three
low dots~($\,\ldots\,$), and ^|\cdots| when you want three vertically centered
dots~($\,\cdots\,$).

In general, it is best to use |\cdots| between $+$ and $-$ and~$\times$
signs, and also between $=$~signs or $\le$~signs or $\subset$ signs or other
similar relations. Low dots are used between ^{commas}, and when things
are juxtaposed with no signs between them at all. For example:
\beginmathdemo
|$x_1+\cdots+x_n$|&x_1+\cdots+x_n\cr
|$x_1=\cdots=x_n=0$|&x_1=\cdots=x_n=0\cr
|$A_1\times\cdots\times A_n$|&A_1\times\cdots\times A_n\cr
|$f(x_1,\ldots,x_n)$|&f(x_1,\ldots,x_n)\cr
|$x_1x_2\ldots x_n$|&x_1x_2\ldots x_n\cr
|$(1-x)(1-x^2)\ldots(1-x^n)$|&(1-x)(1-x^2)\ldots(1-x^n)\cr
|$n(n-1)\ldots(1)$|&n(n-1)\ldots(1)\cr
\endmathdemo

\exercise Type the formulas `$x_1+x_1x_2+\cdots+x_1x_2\ldots x_n$' and
`$(x_1,\ldots,x_n)\cdot(y_1,\ldots,y_n)=x_1y_1+\cdots+x_ny_n$'.
\ [{\sl Hint:\/} A single raised dot is called `^|\cdot|'.]
\answer |$x_1+x_1x_2+\cdots+x_1x_2\ldots x_n$| \ and\hfil\break
|$(x_1,\ldots,x_n)\cdot(y_1,\ldots,y_n)=x_1y_1+\cdots+x_ny_n$|.

But there's an important special case in which |\ldots| and |\cdots|
don't give the correct spacing, namely when they appear at the very end
of a formula, or when they appear just before a closing delimiter like~`|)|'.
In such situations an extra ^{thin space} is needed. For example, consider
sentences like this:
\begindisplay
Prove that $(1-x)^{-1}=1+x+x^2+\cdots\,$.\cr
Clearly $a_i<b_i$ for $i=1$,~2, $\ldots\,$,~$n$.\cr
The coefficients $c_0$,~$c_1$, \dots,~$c_n$ are positive.\cr
\enddisplay
To get the first sentence, the author typed
\begintt
Prove that $(1-x)^{-1}=1+x+x^2+\cdots\,$.
\endtt
Without the `^|\,|' the period would have come too close to the |\cdots|.
Similarly, the second sentence was typed thus:
\begintt
Clearly $a_i<b_i$ for $i=1$,~2, $\ldots\,$,~$n$.
\endtt
Notice the use of ^{ties}, which prevent bad line breaks as explained in
Chapter~14. Such ellipses are extremely common in some forms of mathematical
writing, so plain \TeX\ allows you to say just `^|\dots|' as an abbreviation
for `|$\ldots\,$|' in the text of a paragraph. The third sentence
can therefore be typed
\begintt
The coefficients $c_0$,~$c_1$, \dots,~$c_n$ are positive.
\endtt

\exercise B. C. ^{Dull} tried to take a shortcut by typing the
second example this way:
\begintt
Clearly $a_i<b_i$ for~$i=1, 2, \ldots, n$.
\endtt
What's so bad about that?
\answer The commas belong to the sentence, not to the formula; his
decision to put them into math mode meant that \TeX\ didn't put large
enough spaces after them. Also, his formula `$i=1, 2, \ldots, n$' allows
no breaks between lines, except after the $=$, so he's risking
overfull box problems. But suppose the sentence had been more terse:
\begindisplay
Clearly $a_i<b_i$ \ ($i=1,2,\ldots,n$).
\enddisplay
Then his idea would be basically correct:
\begintt
Clearly $a_i<b_i$ \ ($i=1,2,\ldots,n$).
\endtt

\exercise How do you think the author typed the ^{footnote} in Chapter 4
of this book?
\answer $\ldots$ |never\footnote*{Well \dots, hardly ever.} have| $\ldots$

\subsection Line breaking.  When you have formulas in a paragraph, \TeX\
may have to break them between lines.  This is a necessary evil, something
like the hyphenation of words; we want to avoid it unless the alternative
is worse. ^^{line breaking in math} ^^{breaking formulas between lines}

A formula will be broken only after a relation symbol like $=$
or~$<$ or $\to$, or after a binary operation symbol like $+$ or $-$ or
$\times$, where the relation or binary operation is on the ``outer level'' of
the formula (i.e., not enclosed in |{...}| and not part of an `|\over|'
construction). For example, if you type
\begintt
$f(x,y) = x^2-y^2 = (x+y)(x-y)$
\endtt
in mid-paragraph, there's a chance that \TeX\ will break after either of the
|=|~signs (it prefers this) or after the~|-| or~|+| or~|-| (in an emergency).
But there won't be a break after the comma in any case---commas
after which breaks are desirable shouldn't appear between |$|'s.

If you don't want to permit breaking in this example except after the
|=|~signs, you could type
\begintt
$f(x,y) = {x^2-y^2} = {(x+y)(x-y)}$
\endtt
because these additional braces ``freeze'' the ^{subformulas}, putting them
into unbreakable boxes in which the glue has been set to its natural width.
But it isn't necessary to bother worrying about such things unless \TeX\
actually does break a formula badly, since the chances of this are
pretty slim.

\danger A ``^{discretionary multiplication sign}'' is allowed in formulas:
If you type `|$(x+y)\*(x-y)$|', \TeX\ will treat the ^|\*| something like
the way it treats \hbox{|\-|}; namely, a line break will be allowed at
that place, with the hyphenation penalty. However, instead of inserting a
hyphen, \TeX\ will insert a $\times$ sign in text size.

\danger If you do want to permit a break at some point in the outer level
of a formula, you can say ^|\allowbreak|. For example, if the formula
\begintt
$(x_1,\ldots,x_m,\allowbreak y_1,\ldots,y_n)$
\endtt
appears in the text of a paragraph, \TeX\ will allow it to be broken into the
two pieces `$(x_1,\ldots,x_m,$' and `$y_1,\ldots,y_n)$'.

\ddanger The penalty for breaking after a Rel atom is called ^|\relpenalty|,
and the penalty for breaking after a Bin atom is called ^|\binoppenalty|.
Plain \TeX\ sets |\relpenalty=500| and |\binoppenalty=700|. You can change
the penalty for breaking in any particular case by typing `^|\penalty|\<number>'
immediately after the atom in question; then the number you have specified
will be used instead of the ordinary penalty. For example, you can prohibit
breaking in the formula `$x=0$' by typing `|$x=\nobreak0$|', since
^|\nobreak| is an abbreviation for `|\penalty10000 |'.

\ddangerexercise Is there any difference between the results of
`|$x=\nobreak0$|' and `|${x=0}$|'?
\answer Neither formula will be broken between lines, but the thick spaces
in the second formula will be set to their natural width while the thick
spaces in the first formula will retain their stretchability.

\ddangerexercise How could you prohibit all breaks in formulas, by making only
a few changes to the macros of plain \TeX?
\answer Set ^|\relpenalty||=10000| and ^|\binoppenalty||=10000|.
And you also need to change the definitions of\/ ^|\bmod| and ^|\pmod|,
which insert their own penalties.

\subsection Braces. A variety of different notations have sprung up involving
the symbols `$\{$' and `$\}$'; plain \TeX\ includes several control
sequences that help you cope with formulas involving such things.
^^{braces} ^^{leftbrace} ^^{rightbrace}

In simple situations, braces are used to indicate a ^{set} of objects;
for example, `$\{a,b,c\}$' stands for the set of three objects $a$, $b$,
and~$c$. There's nothing special about typesetting such formulas, except
that you must remember to use |\{| and |\}| for the braces:
^^|\leftbrace| ^^|\rightbrace|
\beginmathdemo
|$\{a,b,c\}$|&\{a,b,c\}\cr
|$\{1,2,\ldots,n\}$|&\{1,2,\ldots,n\}\cr
|$\{\rm red,white,blue\}$|&\{\rm red,white,blue\}\cr
\endmathdemo
A slightly more complex case arises when a set is indicated by giving a
generic element followed by a specific condition; for example, `$\{\,x\mid
x>5\,\}$' stands for the set of all objects $x$ that are greater than~5.
In such situations the control sequence ^|\mid| should be used for the
^{vertical bar}, and thin spaces should be inserted inside the braces:
\beginmathdemo
|$\{\,x\mid x>5\,\}$|&\{\,x\mid x>5\,\}\cr
|$\{\,x:x>5\,\}$|&\{\,x:x>5\,\}\cr
\endmathdemo
(Some authors prefer to use a ^{colon} instead of `$\mid$', as in the second
example here.) \ When the delimiters get larger, as in
\begindisplay
$\displaystyle\bigl\{\,\bigl(x,f(x)\bigr)\bigm\vert x\in D\,\bigr\}$
\enddisplay
they should be called ^|\bigl|, ^|\bigm|, and~^|\bigr|; for example,
the formula just given would be typed
\begintt
\bigl\{\,\bigl(x,f(x)\bigr)\bigm|char`||x\in D\,\bigr\}
\endtt
and formulas that involve still larger delimiters would use ^|\Big| or
^|\bigg| or~even ^|\Bigg|, as explained in Chapter~17.

\exercise How would you typeset the formula
$\bigl\{\,x^3\bigm\vert h(x)\in\{-1,0,+1\}\,\bigr\}$?
\answer |$\bigl\{\,x^3\bigm|\||h(x)\in\{-1,0,+1\}\,\bigr\}$|.

\dangerexercise Sometimes the condition that defines a set is given as
a fairly long English description, not as a formula; for example, consider
`$\{\,p\mid p$~and $p+2$ are prime$\,\}$'. An hbox would do the job:
\begintt
$\{\,p\mid\hbox{$p$ and $p+2$ are prime}\,\}$
\endtt
but a long formula like this is troublesome in a paragraph, since an hbox
cannot be broken between lines, and since the glue inside the |\hbox| does
not vary with the interword glue in the line that contains it. Explain how
the given formula could be typeset with line breaks allowed. [{\sl Hint:\/}
Go back and forth between math ^{mode} and horizontal mode.]
\answer |$\{\,p\mid p$~and $p+2$ are prime$\,\}$|, assuming that
^|\mathsurround| is zero. The more difficult alternative
`|$\{\,p\mid p\ {\rm and}\ p+2\rm\ are\ prime\,\}$|' is not a solution,
because line breaks do not occur at |\|\] ^^|\space|
(or at glue of any kind) within math formulas. Of course it may be best to
display a formula like this, instead of breaking it between lines.

Displayed formulas often involve another sort of brace, to indicate a choice
between various alternatives, as in the construction
\begindisplay
$\displaystyle\vert x\vert=\cases{x,&if $x\ge0$;\cr -x,&otherwise.\cr}$
\enddisplay
^^{selection, see cases} ^^{alternatives, see cases} ^^{choices, see cases}
You can typeset it with the control sequence ^|\cases|:
\begintt
$$|char`||x||=\cases{x,&if $x\ge0$;\cr
             -x,&otherwise.\cr}$$
\endtt
Look closely at this example and notice that it uses the character |&|,
^^{ampersand} which we said in Chapter~7 was reserved for special purposes.
Here for the first time in this manual we have an example of why |&|~is
so special: Each of the cases has two parts, and the~|&| separates those
parts. To the left of the~|&| is a math formula that is implicitly
enclosed in |$...$|; to the right of the~|&| is ordinary text, which is
{\sl not\/} implicitly enclosed in |$...$|. For example, the `|-x,|' in
the second line will be typeset in math mode, but the `|otherwise|' will
be typeset in horizontal mode. Blank spaces after the~|&| are ignored.
There can be any number of cases, but there usually are at least two.
Each case should be followed by ^|\cr|. Notice that the |\cases| construction
typesets its own `$\{$'; there is no corresponding `$\}$'.

\exercise Typeset the display \lower12pt\null\
$\smash{\displaystyle
f(x)=\cases{1/3&if $0\le x\le1$;\cr 2/3&if $3\le x\le4$;\cr 0&elsewhere.\cr}
}$
\answer |$$f(x)=\cases{1/3&if $0\le x\le1$;\cr 2/3&if $3\le x\le4$;\cr|\hfil
\break|0&elsewhere.\cr}$$|

\danger You can insert `^|\noalign||{|$\langle$vertical mode
material$\rangle$|}|' just after any \kern-1pt|\cr| within |\cases|, as
explained in Chapter~22, because |\cases| is an application of the general
alignment constructions considered in that chapter. For example, the
command `|\noalign{\vskip2pt}|' can be used to put a little extra space
between two of the cases.

\danger ^{Horizontal braces} will be set over or under parts of a displayed
formula if you use the control sequences ^|\overbrace| or ^|\underbrace|.
Such constructions are considered to be large operators like |\sum|, so you
can put limits above them or below them by specifying superscripts or
subscripts, as in the following examples:
\beginlongdisplaymathdemo
\noalign{\vskip9pt}
|$$\overbrace{x+\cdots+x}^{k\rm\;times}$$|&
  \overbrace{x+\cdots+x}^{k\rm\;times}\cr
\noalign{\vskip-6pt}
|$$\underbrace{x+y+z}_{>\,0}.$$|&
  \underbrace{x+y+z}_{>\,0}.\cr
\endmathdemo

\subsection Matrices. Now comes the fun part. Mathematicians in many different
disciplines like to construct rectangular arrays of formulas that have been
arranged in rows and columns; such an ^{array} is called a {\sl^{matrix}}.
Plain \TeX\ provides a ^|\matrix| control sequence that makes it convenient
to deal with the most common types of matrices.

For example, suppose that you want to specify the display
$$A=\left(\matrix{x-\lambda&1&0\cr
                  0&x-\lambda&1\cr
                  0&0&x-\lambda\cr}\right).$$
All you do is type
\begintt
$$A=\left(\matrix{x-\lambda&1&0\cr
                  0&x-\lambda&1\cr
                  0&0&x-\lambda\cr}\right).$$
\endtt
^^|\lambda|
This is very much like the |\cases| construction we looked at earlier;
each row of the matrix is followed by~|\cr|, and `|&|'~signs are used
between the individual entries of each row. Notice, however, that you are
supposed to put your own |\left| and |\right| delimiters around the matrix;
this makes |\matrix| different from |\cases|, which inserts a big `$\{$'
automatically.  The reason is that |\cases| always involves a left brace,
but different delimiters are used in different matrix constructions. On
the other hand, parentheses are used more often than other delimiters, so
you can write ^|\pmatrix| if you want plain \TeX\ to fill in the
parentheses for you; the example above then reduces to
\begintt
$$A=\pmatrix{x-\lambda&...&x-\lambda\cr}.$$
\endtt

\dangerexercise Typeset the display \ \lower12pt\null
$\tenpoint\smash{\displaystyle
\left\lgroup\matrix{a&b&c\cr d&e&f\cr}\right\rgroup
  \left\lgroup\matrix{u&x\cr v&y\cr w&z\cr}\right\rgroup
}$, \
using ^|\lgroup| and ^|\rgroup|.
\answer |$$\left\lgroup\matrix{a&b&c\cr d&e&f\cr}\right\rgroup|\hfil\break
|\left\lgroup\matrix{u&x\cr v&y\cr w&z\cr}\right\rgroup$$|.

\danger The individual entries of a matrix are normally centered in columns.
Each column is made as wide as necessary to accommodate the entries it
contains, and there's a ^{quad} of space between columns.
If you want to put something ^{flush right} in its column, precede it
by ^|\hfill|; if you want to put something ^{flush left} in its column,
follow it by~|\hfill|.

\danger Each entry of a matrix is treated separately from the others,
and it is typeset as a math formula in text style. Thus, for example,
if you say |\rm| in one entry, it does not affect the others.
Don't try to say `|{\rm x&y}|'.

Matrices often appear in the form of generic patterns that use ^{ellipses}
(i.e., dots) to indicate rows or columns that are left out. You can typeset
such matrices by putting the ellipses into rows and/or columns of their own.
Plain \TeX\ provides ^|\vdots| (vertical dots) and ^|\ddots| (diagonal dots)
as companions to ^|\ldots| for constructions like this. For example, the
^{generic matrix}
$$A=\pmatrix{a_{11}&a_{12}&\ldots&a_{1n}\cr
             a_{21}&a_{22}&\ldots&a_{2n}\cr
             \vdots&\vdots&\ddots&\vdots\cr
             a_{m1}&a_{m2}&\ldots&a_{mn}\cr}$$
is easily specified:
\begintt
$$A=\pmatrix{a_{11}&a_{12}&\ldots&a_{1n}\cr
             a_{21}&a_{22}&\ldots&a_{2n}\cr
             \vdots&\vdots&\ddots&\vdots\cr
             a_{m1}&a_{m2}&\ldots&a_{mn}\cr}$$
\endtt

\medskip
\exercise How can you get \TeX\ to produce the ^{column vector} ^^{vector}
\lower18pt\null\ $\smash{\displaystyle
\pmatrix{y_1\cr \vdots\cr y_k\cr}
}$\quad?
\answer |\pmatrix{y_1\cr \vdots\cr y_k\cr}|.

\danger Sometimes a matrix is bordered at the top and left by formulas
that give labels to the rows and columns. Plain \TeX\ provides a special
macro called ^|\bordermatrix| for this situation. For example, the display
$$\tenmath
M=\bordermatrix{&C&I&C'\cr C&1&0&0\cr I&b&1-b&0\cr C'&0&a&1-a\cr}$$
is obtained when you type
\begintt
$$M=\bordermatrix{&C&I&C'\cr
                C&1&0&0\cr  I&b&1-b&0\cr  C'&0&a&1-a\cr}$$
\endtt
The first row gives the upper labels, which appear above the big left
and right parentheses; the first column gives the left labels, which are
typeset flush left, just before the matrix itself. The first column in
the first row is normally blank. Notice that |\bordermatrix| inserts
its own parentheses, like |\pmatrix| does.

\danger It's usually inadvisable to put matrices into the text of a paragraph,
because they are so big that they are better displayed. But occasionally
you may want to specify a small matrix like $1\,1\choose0\,1$, which you can
^^|\choose| ^^{matrix, small}
typeset for example as `|$1\,1\choose0\,1$|'. Similarly, the small matrix
$\bigl({a\atop l}{b\atop m}{c\atop n}\bigr)$ can be typeset as
\begintt
$\bigl({a\atop l}{b\atop m}{c\atop n}\bigr)$
\endtt
^^|\atop| The |\matrix| macro does not produce small arrays of this sort.

\subsection Vertical spacing. If you want to tidy up an unusual formula,
you know already how to move things farther apart or closer together, by
using positive or negative thin spaces. But such spaces affect only the
horizontal dimension; what if you want something to be moved higher
or lower? That's an advanced topic.

\danger Appendix B provides a few macros that can be used to fool \TeX\
into thinking that certain formulas are larger or smaller than they really
are; such tricks can be used to move other parts of the formula up or down
or left or right. For example, we have already discussed the use of
^|\mathstrut| in Chapter~16 and ^|\strut| in Chapter~17; these invisible
boxes caused \TeX\ to put square root signs and the denominators of
continued fractions into different positions than usual.

\danger If you say `^|\phantom||{|\<subformula>|}|' in any formula, plain
\TeX\ will do its spacing as if you had said simply
`|{|\<subformula>|}|', but the subformula itself will be invisible. Thus,
for example, `|\phantom{0}2|' takes up just as much space as `|02|' in the
current style, but only the~|2| will actually appear on the page. If you
want to leave blank space for a ^{new symbol} that has exactly the same
size as $\sum$, but if you are forced to put that symbol in by hand for
some reason, `|\mathop{\phantom\sum}|' will leave exactly the right amount
of blank space.  \ (The `^|\mathop|' here makes this phantom behave like
|\sum|, i.e., as a large operator.)

\danger Even more useful than |\phantom| is ^|\vphantom|, which makes
an invisible box whose height and depth are the same as those of
the corresponding |\phantom|, but
the width is zero. Thus, |\vphantom| makes a vertical ^{strut} that can
increase a formula's effective height or depth. Plain \TeX\ defines
|\mathstrut| to be an abbreviation for `|\vphantom(|'. There's also
^|\hphantom|, which has the width of a |\phantom|, but its height
and depth are zero.

\danger Plain \TeX\ also provides `^|\smash||{|\<subformula>|}|', a macro
that yields the same result as `|{|\<subformula>|}|' but makes the height and
depth zero. By using both |\smash| and |\vphantom| you can typeset any
subformula and give it any desired nonnegative height and depth. For example,
\begintt
\mathop{\smash\limsup\vphantom\liminf}
\endtt
produces a large operator that says `$\limsup$', but its height and depth
are those of\/ ^|\liminf| (i.e., the depth is zero). ^^|\limsup|

\def\undertext#1{$\underline{\hbox{#1}}$}
\ddangerexercise If you want to underline some text, you could use a macro like
\begintt
\def\undertext#1{$\underline{\hbox{#1}}$}
\endtt
to do the job. \undertext{But} \undertext{this} \undertext{doesn't}
\undertext{always} \undertext{work} \undertext{right}. Discuss better
alternatives. ^^{underlined text}
\answer |\def|\stretch|\undertext|\stretch|#1{$\underline|\stretch
|{\smash|\stretch|{\hbox|\stretch|{#1}}}$}| will underline the
\def\undertext#1{$\underline{\smash{\hbox{#1}}}$}%
words and cross \undertext{through} the descenders; or you could insert
|\vphantom{y}| before the |\hbox|, thereby lowering all of the underlines
to a position below all descenders. Neither of these gives exactly what is
wanted. \ (See also ^|\underbar| in Appendix~B\null.) \ Underlining is actually
not very common in fine typography, since font changes usually work just
as well or better, when you want to emphasize something. If you really want
underlined text, it's best to have a special font in which all
the letters are underlined.

\ddanger You can also use ^|\raise| and ^|\lower| to adjust the vertical
positions of boxes in formulas. For example, the formula
`|$2^{\raise1pt\hbox{$\scriptstyle n$}}$|' will have its superscript~$n$
one point higher than usual ($2^{\raise1pt\hbox{$\scriptstyle n$}}$ instead of
$2^n$). Note that it was necessary to say ^|\scriptstyle| in this example,
since the contents of an ^|\hbox| will normally be in text style even when
that hbox appears in a superscript, and since |\raise| can be used only in
connection with a box.  This method of positioning is not used extremely
often, but it is sometimes helpful if the ^|\root| macro doesn't put its
argument in a suitable place. For example,
\begindisplay
|\root\raise|\<dimen>|\hbox{$\scriptscriptstyle|\<argument>|$}\of...|
\enddisplay
will move the argument up by a given amount.

\ddanger Instead of changing the sizes of subformulas, or using |\raise|,
you can also control vertical spacing by changing the parameters
that \TeX\ uses when it is converting math lists to horizontal lists.
These parameters are described in Appendix~G\null; you need to be careful when
changing them, because such changes are ^{global} (i.e., not local to groups).
Here is an example of how such a change might be made: Suppose that you
are designing a format for ^{chemical typesetting}, and that you expect to be
setting a lot of formulas like `$\rm Fe_2^{+2}Cr_2O_4$'. You may not like the
fact that the subscript in~$\rm Fe_2^{+2}$ is lower than the subscript
in~$\rm Cr_2$; and you don't want to force users to type monstrosities like
\begintt
$\rm Fe_2^{+2}Cr_2^{\vphantom{+2}}O_4^{\vphantom{+2}}$
\endtt
just to get the formula
$\rm Fe_2^{+2}Cr_2^{\vphantom{+2}}O_4^{\vphantom{+2}}$
with all subscripts at the same level. Well, all you need to do is set
`^|\fontdimen||16\tensy=2.7pt|' and `|\fontdimen17\tensy=2.7pt|', assuming
that ^|\tensy| is your main symbol font (|\textfont2|); this lowers all
normal ^{subscripts} to a position $2.7\pt$ below the baseline, which is
enough to make room for a possible superscript that contains a plus sign.
Similarly, you can adjust the positioning of ^{superscripts} by changing
|\fontdimen14\tensy|. There are parameters for the position of the ^{axis line},
the positions of ^{numerator} and ^{denominator} in a generalized ^{fraction},
the spacing above and below ^{limits}, the default ^{rule thickness}, and so
on. Appendix~G gives precise details.

\subsection Special features for math hackers. \TeX\ has a few more primitive
operations for math mode that haven't been mentioned yet. They are
occasionally useful if you are designing special formats.

\ddanger If a glue or kern specification is immediately preceded by
`^|\nonscript|', \TeX\ will not use that glue or kern in
script or scriptscript styles. Thus, for example, the sequence
`|\nonscript\;|' produces exactly the amount of space specified by
`{\tt(3)}' in the spacing table for mathematics that appeared earlier
in this chapter.

\ddanger Whenever \TeX\ has scanned a |$| and is about to read a math formula
that appears in text, it will first read another list of tokens that has
been predefined by the command ^|\everymath||={|\<token list>|}|. \ (This is
analogous to |\everypar|, which was described in Chapter~14.) \ Similarly,
you can say ^|\everydisplay||={|\<token list>|}| to predefine a list of tokens
for \TeX\ to read just after it has scanned an opening |$$|, i.e., just
before reading a formula that is to be displayed. With |\everymath| and
|\everydisplay|, you can set up special conventions that you wish to apply
to all formulas.

\subsection Summary. We have discussed more different kinds of formulas in
this chapter than you will usually find in any one book of mathematics.
If you have faithfully done the exercises so far, you can face almost
any formula with confidence.

\newcount\chalcount \chalcount=0
\outer\def\challenge{\d@nger\chall}
\outer\def\cchallenge{\dd@nger\chall}
\def\chall{\global\advance\chalcount by1
  \dexercise \hbox{Challenge number \the\chalcount:\enskip}\ignorespaces}

\danger But here are a few more exercises, to help you review what you
have learned. Each of the following ``challenge formulas'' illustrates one
or more of the principles already discussed in this chapter. The author
confesses that he is trying to trip you~up on several of these.
Nevertheless, if you try each one before looking at the answer, and if
you're alert for traps, you should find that these formulas provide a good
way to consolidate and complete your knowledge.

\challenge Explain how to type the phrase `$n^{\rm th}$ root', where
`$n^{\rm th}$' is treated as a mathematical formula with a superscript in
roman type.
\answer |$n^{\rm th}$ root|. \ (Incidentally, it is also acceptable
to type `|$n$th|', getting `$n$th', in such situations; the fact that
the $n$ is in italics distinguishes it from the suffix. Typed manuscripts
generally render this with a hyphen, but `$n$-th' is frowned on nowadays
when an italic~$n$ is available.) ^^{nth}

\challenge $\qquad\tenmath{\bf S^{\rm-1}TS=dg}(\omega_1,\ldots,\omega_n)
=\bf\Lambda$.
\answer |${\bf S^{\rm-1}TS=dg}(\omega_1,|\stretch|\ldots,|\stretch|\omega_n)
=\bf\Lambda$|.
\ $\bigl($Did you notice the difference between ^|\omega| ($\omega$)
and~|w| ($w$)?$\bigr)$

\challenge $\qquad\tenmath\Pr(\,m=n\mid m+n=3\,)$.
\answer |$\Pr(\,m=n\mid m+n=3\,)$|. \ (Analogous to a set.) ^^|\Pr|

\challenge $\qquad\tenmath\sin18^\circ={1\over4}(\sqrt5-1)$.^^{degrees}
\answer |$\sin18^\circ={1\over4}(\sqrt5-1)$|. ^^|\circ|

\challenge $\qquad\tenmath k=1.38065\times10^{-16}\rm\,erg\,K^{-1}$.
\answer |$k=1.38065\times10^{-16}\rm\,erg\,K^{-1}$|.

\challenge $\qquad\tenmath \bar\Phi\subset NL_1^*/N=\bar L_1^*
  \subseteq\cdots\subseteq NL_n^*/N=\bar L_n^*$.
\answer |$\bar\Phi\subset NL_1^*/N=\bar L_1^*|\parbreak
        |  \subseteq\cdots\subseteq NL_n^*/N=\bar L_n^*$|.

\challenge $\qquad\tenmath I(\lambda)=\int\!\!\int_Dg(x,y)e^{i\lambda h(x,y)}
  \,dx\,dy$. % cf. Math. Comp. 37 (1981), 509
\answer |$I(\lambda)=\int\!\!\int_Dg(x,y)e^{i\lambda h(x,y)}\,dx\,dy$|.\hfil
\break
(Although three |\!|'s work out best between consecutive integral signs in
displays, the text style seems to want only two.) ^^{double integral}
^^{integral, multiple}

\challenge $\qquad\tenmath
  \int_0^1\!\cdots\int_0^1f(x_1,\ldots,x_n)\,dx_1\ldots\,dx_n$.
\answer |$\int_0^1\!\cdots\int_0^1f(x_1,\ldots,x_n)\,dx_1\ldots\,dx_n$|.

\challenge Here's a display.
$$\tenmath x_{2m}\equiv\cases{Q(X_m^2-P_2W_m^2)-2S^2&($m$ odd)\cr
  \noalign{\vskip2pt}
  P_2^2(X_m^2-P_2W_m^2)-2S^2&($m$ even)\cr}\pmod N.$$
\answer |$$x_{2m}\equiv\cases{Q(X_m^2-P_2W_m^2)-2S^2&($m$ odd)\cr|\parbreak
        |       \noalign{\vskip2pt} % spread the lines apart a little|\parbreak
        |       P_2^2(X_m^2-P_2W_m^2)-2S^2&($m$ even)\cr}\pmod N.$$|

\challenge And another. % ACP Eq. 1.2.9--33
$$\tenmath (1+x_1z+x_1^2z^2+\cdots\,)\ldots(1+x_nz+x_n^2z^2+\cdots\,)
  ={1\over(1-x_1z)\ldots(1-x_nz)}.$$
\answer |$$(1+x_1z+x_1^2z^2+\cdots\,)\ldots(1+x_nz+x_n^2z^2+\cdots\,)|\parbreak
        |  ={1\over(1-x_1z)\ldots(1-x_nz)}.$$| \ (Notice the uses of\/ |\,|.)

\challenge And another. % Eq. 1.2.9--9
$$\tenmath \prod_{j\ge0}\biggl(\sum_{k\ge0}a_{jk}z^k\biggr)
  =\sum_{n\ge0}z^n\,\Biggl(\sum_
     {\scriptstyle k_0,k_1,\ldots\ge0\atop
      \scriptstyle k_0+k_1+\cdots=n}
   a_{0k_0}a_{1k_1}\ldots\,\Biggr).$$
\answer |$$\prod_{j\ge0}\biggl(\sum_{k\ge0}a_{jk}z^k\biggr)|\parbreak
        |  =\sum_{n\ge0}z^n\,\Biggl(\sum_|\parbreak
        |     {\scriptstyle k_0,k_1,\ldots\ge0\atop|\parbreak
        |      \scriptstyle k_0+k_1+\cdots=n}|\parbreak
        |   a_{0k_0}a_{1k_1}\ldots\,\Biggr).$$|\par
\nobreak\smallskip\noindent Some people would prefer to have the latter
parentheses larger; but |\left| and |\right| come out a bit too large in this
case. It's not difficult to define ^|\bigggl| and ^|\bigggr| macros, analogous
to the definitions of\/ |\biggl| and |\biggr| in Appendix~B.

\challenge And, % cf ACP vol1 p64
$$\tenmath {(n_1+n_2+\cdots+n_m)!\over n_1!\,n_2!\ldots n_m!}
  ={n_1+n_2\choose n_2}{n_1+n_2+n_3\choose n_3}
    \ldots{n_1+n_2+\cdots+n_m\choose n_m}.$$
\answer |$${(n_1+n_2+\cdots+n_m)!\over n_1!\,n_2!\ldots n_m!}|\parbreak
        |  ={n_1+n_2\choose n_2}{n_1+n_2+n_3\choose n_3}|\parbreak
        |    \ldots{n_1+n_2+\cdots+n_m\choose n_m}.$$|

\challenge Yet another display. % found in Chaundy et al
$$\tenmath \def\\#1#2{(1-q^{#1_#2+n})} % to save typing
\Pi_R{a_1,a_2,\ldots,a_M\atopwithdelims[]b_1,b_2,\ldots,b_N}
  =\prod_{n=0}^R{\\a1\\a2\ldots\\aM\over\\b1\\b2\ldots\\bN}.$$
\answer |$$\def\\#1#2{(1-q^{#1_#2+n})} % to save typing|\parbreak
        |\Pi_R{a_1,a_2,\ldots,a_M\atopwithdelims[]b_1,b_2,\ldots,b_N}|\parbreak
        |  =\prod_{n=0}^R{\\a1\\a2\ldots\\aM\over\\b1\\b2\ldots\\bN}.$$|
^^|\atopwithdelims|

\challenge And another.
$$\tenmath \sum_{p\rm\;prime}f(p)=\int_{t>1}f(t)\,d\pi(t).$$
\answer |$$\sum_{p\rm\;prime}f(p)=\int_{t>1}f(t)\,d\pi(t).$$|

\challenge Still another.
$$\tenmath \{\underbrace{\overbrace{\mathstrut a,\ldots,a}
      ^{k\;a\mathchar`'\rm s},
    \overbrace{\mathstrut b,\ldots,b}
      ^{l\;b\mathchar`'\rm s}}_{k+l\rm\;elements}\}.$$
\answer |$$\{\underbrace{\overbrace{\mathstrut a,\ldots,a}|\parbreak
        |      ^{k\;a\mathchar`'\rm s},|\parbreak
        |    \overbrace{\mathstrut b,\ldots,b}|\parbreak
        |      ^{l\;b\mathchar`'\rm s}}_{k+l\rm\;elements}\}.$$|\par
\smallskip\noindent Notice how ^{apostrophes} (instead of primes) were obtained.

\challenge Put a ^|\smallskip| between the rows of matrices in the
compound matrix ^^{compound matrix}
$$\tenmath \pmatrix{\pmatrix{a&b\cr c&d\cr}&
             \pmatrix{e&f\cr g&h\cr}\cr
           \noalign{\smallskip}
           0&\pmatrix{i&j\cr k&l\cr}\cr}.$$
\answer |$$\pmatrix{\pmatrix{a&b\cr c&d\cr}&|\parbreak
        |             \pmatrix{e&f\cr g&h\cr}\cr|\parbreak
        |           \noalign{\smallskip}|\parbreak
        |           0&\pmatrix{i&j\cr k&l\cr}\cr}.$$|

\challenge Make the columns ^{flush left} here. % cf Polya/Szego VII.43.2
$$\tenmath \det\left\vert\,\matrix{
  c_0&c_1\hfill&c_2\hfill&\ldots&c_n\hfill\cr
  c_1&c_2\hfill&c_3\hfill&\ldots&c_{n+1}\hfill\cr
  c_2&c_3\hfill&c_4\hfill&\ldots&c_{n+2}\hfill\cr
  \,\vdots\hfill&\,\vdots\hfill&
       \,\vdots\hfill&&\,\vdots\hfill\cr
  c_n&c_{n+1}\hfill&c_{n+2}\hfill&\ldots&c_{2n}\hfill\cr
  }\right\vert>0.$$
\answer |$$\det\left|\||\,\matrix{|\parbreak
        |  c_0&c_1\hfill&c_2\hfill&\ldots&c_n\hfill\cr|\parbreak
        |  c_1&c_2\hfill&c_3\hfill&\ldots&c_{n+1}\hfill\cr|\parbreak
        |  c_2&c_3\hfill&c_4\hfill&\ldots&c_{n+2}\hfill\cr|\parbreak
        |  \,\vdots\hfill&\,\vdots\hfill&|\parbreak
        |       \,\vdots\hfill&&\,\vdots\hfill\cr|\parbreak
        |  c_n&c_{n+1}\hfill&c_{n+2}\hfill&\ldots&c_{2n}\hfill\cr|\parbreak
        |  }\right|\||>0.$$|

\cchallenge The main problem here is to prime the $\sum$.
^^|\sum prime| ^^{=def}
$$\tenmath \mathop{{\sum}'}_{x\in A}f(x)\mathrel{\mathop=^{\rm def}}
  \sum_{\scriptstyle x\in A\atop\scriptstyle x\ne0}f(x).$$
\answer |$$\mathop{{\sum}'}_{x\in A}f(x)\mathrel{\mathop=^{\rm def}}|\parbreak
        |  \sum_{\scriptstyle x\in A\atop\scriptstyle x\ne0}f(x).$$|\par
\smallskip\noindent
This works because |{\sum}| is type Ord (so its superscript is not set
above), but ^|\mathop||{{\sum}'}| is type Op (so its subscript is set below).
The limits are centered on $\sum'$, however, not on $\sum$. If you don't
like that, the remedy is more difficult; one solution is to use
|\sumprime_{x\in A}| where ^|\sumprime| is defined as follows:
\par\nobreak\medskip
|\def\sumprime_#1{\setbox0=\hbox{$\scriptstyle{#1}$}|\parbreak
|  \setbox2=\hbox{$\displaystyle{\sum}$}|\parbreak
|  \setbox4=\hbox{${}'\mathsurround=0pt$}|\parbreak
|  \dimen0=.5\wd0 \advance\dimen0 by-.5\wd2|\parbreak
|  \ifdim\dimen0>0pt|\parbreak
|    \ifdim\dimen0>\wd4 \kern\wd4 \else\kern\dimen0\fi\fi|\parbreak
|  \mathop{{\sum}'}_{\kern-\wd4 #1}}|

\cchallenge You may be ready now for this display.
$$\tenmath 2\uparrow\uparrow k\mathrel{\mathop=^{\rm def}}
  2^{2^{2^{\cdot^{\cdot^{\cdot^2}}}}}
    \vbox{\hbox{$\Big\}\scriptstyle k$}\kern0pt}.$$
\answer |$$2\uparrow\uparrow k\mathrel{\mathop=^{\rm def}}|\parbreak
        |  2^{2^{2^{\cdot^{\cdot^{\cdot^2}}}}}|\parbreak
        |    \vbox{\hbox{$\Big\}\scriptstyle k$}\kern0pt}.$$|\par

\cchallenge And finally, when you have polished off all the other examples,
here's the ultimate test. Explain how to obtain the ^{commutative diagram}
% from Invent. Math. 70 (1982), 34 % with "typo" corrected 99.12.03
$$\tenmath\def\normalbaselines{\baselineskip20pt\lineskip1pt\lineskiplimit0pt }
\def\mapright#1{\smash{
    \mathop{\longrightarrow}\limits^{#1}}}
\def\mapdown#1{\Big\downarrow
  \rlap{$\vcenter{\hbox{$\scriptstyle#1$}}$}}
\matrix{\noalign{\vskip6pt}&&&&&&0\cr
  &&&&&&\mapdown{}\cr
  0&\mapright{}&{\cal O}_C&\mapright\iota&
    \cal E&\mapright\rho&\cal L&\mapright{}&0\cr
  &&\Big\Vert&&\mapdown\phi&&\mapdown\psi\cr
  0&\mapright{}&{\cal O}_C&\mapright\pi&
    \pi_*{\cal O}_D&\mapright\delta&
    R^1f_*{\cal O}_V(-D)&\mapright{}&0\cr
  &&&&&&\mapdown{\theta_i\otimes\gamma^{-1}}\cr
  &&&&&&\hidewidth R^1f_*\bigl({\cal O}
    _V(-iM)\bigr)\otimes\gamma^{-1}\hidewidth\cr
  &&&&&&\mapdown{}\cr
  &&&&&&0\cr\noalign{\vskip6pt}}$$
using ^|\matrix|. \ (Many of the entries are blank.)
\answer If you have to do a lot of commutative diagrams, you will want to
define some macros like those in the first few lines of this solution.
The ^|\matrix| macro resets the baselines to ^|\normalbaselines|, because
other commands like |\openup| might have changed them, so
we redefine |\normalbaselines| in this solution. Some of the things
shown here haven't been explained yet, but Chapter~22 will reveal all.
\smallskip
|$$\def\normalbaselines{\baselineskip20pt|\parbreak
|  \lineskip3pt \lineskiplimit3pt }|\parbreak
|\def\mapright#1{\smash{|\parbreak
|    \mathop{\longrightarrow}\limits^{#1}}}|\parbreak
|\def\mapdown#1{\Big\downarrow|\parbreak
|  \rlap{$\vcenter{\hbox{$\scriptstyle#1$}}$}}|\parbreak
|\matrix{&&&&&&0\cr|\parbreak
|  &&&&&&\mapdown{}\cr|\parbreak
|  0&\mapright{}&{\cal O}_C&\mapright\iota&|\parbreak
|    \cal E&\mapright\rho&\cal L&\mapright{}&0\cr|\parbreak
|  &&\Big\Vert&&\mapdown\phi&&\mapdown\psi\cr|\parbreak
|  0&\mapright{}&{\cal O}_C&\mapright\pi&|\parbreak
|    \pi_*{\cal O}_D&\mapright\delta&|\parbreak
|    R^1f_*{\cal O}_V(-D)&\mapright{}&0\cr|\parbreak
|  &&&&&&\mapdown{\theta_i\otimes\gamma^{-1}}\cr|\parbreak
|  &&&&&&\hidewidth R^1f_*\bigl({\cal O}|\parbreak
|    _V(-iM)\bigr)\otimes\gamma^{-1}|^|\hidewidth||\cr|\parbreak
|  &&&&&&\mapdown{}\cr|\parbreak
|  &&&&&&0\cr}$$|

\subsection Words of advice. The number of different notations is
enormous and still growing, so you will probably continue to find
new challenges as you continue to type mathematical papers. It's a
good idea to keep a personal notebook in which you record all of
the nonobvious formulas that you have handled successfully,
showing both the final output and what you typed to get it.
Then you'll be able to refer back to those solutions when you
discover that you need to do something similar, a few months later.

If you're a mathematician who types your own papers, you have now learned
^^{author, typesetting by}
how to get enormously complex formulas into print, and you can do so
without going through an intermediary who may somehow distort their
meaning. But please, don't get too carried away by your newfound talent;
the fact that you are able to typeset your formulas with \TeX\ doesn't
necessarily mean that you have found the best notation for communicating
with the readers of your work.  Some notations will be unfortunate even
when they are beautifully formatted.

\endchapter

Mathematicians are like Frenchmen:\/
% Die Mathematiker sind eine Art Franzosen:
whenever you say something to them, they translate it into their own language,
% redet man zu ihnen, so u"bersetzen sie es in ihre Sprache,
and at once it is something entirely different.
% und dann ist es alsobald ganz etwas Anders.
\author ^{GOETHE}, {\sl Maxims and Reflexions\/} (1829)
% see Schriften der Goethe-Gesellschaft, vol 21, pp 266 and 389

\bigskip

The best notation is no notation;
whenever it is possible to avoid the use of a complicated alphabetic apparatus,
avoid it.
A good attitude to the preparation of written mathematical exposition
is to pretend that it is spoken.
Pretend that you are explaining the subject to a friend
on a long walk in the woods, with no paper available;
fall back on symbolism only when it is really necessary.
\author PAUL ^{HALMOS}, {\sl How to Write Mathematics\/} (1970)
  % in {\sl L'Enseignement Math\'ematique\/}
  % vol 16, 123--152; section 15; reprinted in AMS pub "How to Write Math"

\eject
\beginchapter Chapter 19. Displayed Equations

By now you know how to type mathematical formulas so that \TeX\ will handle
them with supreme elegance; your knowledge of math typing is nearly complete.
But there is one more part to the story, and the purpose of this chapter
is to present the happy ending. We have discussed how to deal with individual
formulas; but ^{displays} often involve a whole bunch of different formulas,
or different pieces of a huge formula, and it's a bit of a problem to
lay them out so that they line up properly with each other. Fortunately,
large displays generally fall into a few simple patterns.

\subsection One-line displays. Before plunging into the general question
of display layout, let's recapitulate what we have already covered.  If
you type `|$$|\<formula>|$$|', \TeX\ will display the formula
^^{dollardollar} in flamboyant display style, centering it on a line by
itself. We have also noted in Chapter~18 that it's possible to display two
short formulas at once, by typing
`|$$|\<formula$_1$>^|\qquad|\<formula$_2$>|$$|'; this reduces the
two-formula problem to a one-formula problem. You get the two formulas
separated by two quads of space, the whole being centered on a line.

Displayed equations often involve ordinary text. Chapter~18 explains
how to get roman type into formulas without leaving math mode, but the
best way to get text into a display is to put it into an ^|\hbox|.
There needn't even be any math at all; to typeset
$$\hbox{Displayed Text}$$
you can simply say `|$$\hbox{Displayed Text}$$|'. But here's a more interesting
example:
$$X_n=X_k \qquad\hbox{if and only if}\qquad
    Y_n=Y_k \quad\hbox{and}\quad Z_n=Z_k.$$
Formulas and text were combined in this case by typing
\begintt
$$X_n=X_k \qquad\hbox{if and only if}\qquad
    Y_n=Y_k \quad\hbox{and}\quad Z_n=Z_k.$$
\endtt
Notice that |\qquad| appears around `if and only if', but a single ^|\quad|
surrounds `and'; this helps to indicate that the $Y$ and~$Z$ parts of the
display are related more closely to each other than to the $X$~part.

Consider now the display
$$Y_n=X_n\bmod p \quad\hbox{and}\quad Z_n=X_n\bmod q
    \qquad\hbox{for all }n\ge0.$$
Can you figure out how to type this? One solution is
\begintt
$$Y_n=X_n\bmod p \quad\hbox{and}\quad Z_n=X_n\bmod q
    \qquad\hbox{for all }n\ge0.$$
\endtt
Notice that a space has been left after `|all|' in the hbox here, since spaces
disappear when they are out in formula-land. But there's a simpler and more
logical way to proceed, once you get used to \TeX's idea of modes: You can type
\begintt
... \qquad\hbox{for all $n\ge0$.}$$
\endtt
Wow---that's math mode inside of horizontal mode inside of display
math mode. But in this way your manuscript mirrors what you are trying to
accomplish, while the previous solution (with the space after `|all|')
looks somewhat forced.

\exercise Typeset the following four displays (one at a time):
$$\openup1\jot\displaylines{
\sum_{n=0}^\infty a_nz^n\qquad\hbox{converges if}\qquad
  \vert z\vert<\Bigl(\limsup_{n\to\infty}
    \root n\!\of{\vert a_n\vert}\,\Bigr)^{-1}.\cr
{f(x+\Delta x)-f(x)\over\Delta x}\to f'(x)\qquad\hbox{as $\Delta x\to0$.}\cr
\noalign{\vskip2pt}
\Vert u_i\Vert=1,\qquad u_i\cdot u_j=0\quad\hbox{if $i\ne j$.}\cr
\it\hbox{The confluent image of}\quad\left\{
  \matrix{\hbox{an arc}\hfill\cr\hbox{a circle}\hfill\cr
    \hbox{a fan}\hfill\cr}
  \right\}\quad\hbox{is}\quad\left\{
  \matrix{\hbox{an arc}\hfill\cr\hbox{an arc or a circle}\hfill\cr
    \hbox{a fan or an arc}\hfill\cr}\right\}.\cr
\noalign{\vskip-8pt}}$$
^^|\Delta| ^^|\Vert|
% the last example comes from Proc AMS 55 (1976), 410, with typos corrected
\answer |$$\sum_{n=0}^\infty a_nz^n\qquad\hbox{converges if}\qquad|\parbreak
|  |\||z|\||<\Bigl(\limsup_{n\to\infty}\root n\!\of{|\||a_n|\|^^|\root|
  |}\,\Bigr)^{-1}.$$|\kern-.33pt\par
\smallskip
|$${f(x+\Delta x)-f(x)\over\Delta x}\to f'(x)|\parbreak
|    \qquad\hbox{as $\Delta x\to0$.}$$|\par
\smallskip
|$$\|\||u_i\|\||=1,\qquad u_i\cdot u_j=0\quad\hbox{if $i\ne j$.}$$|\par
\smallskip
|$$\it\hbox{The confluent image of}\quad\left\{|\parbreak
|    \matrix{\hbox{an arc}\hfill\cr\hbox{a circle}\hfill\cr|\parbreak
|      \hbox{a fan}\hfill\cr}|\parbreak
|    \right\}\quad\hbox{is}\quad\left\{|\parbreak
|    \matrix{\hbox{an arc}\hfill\cr|\parbreak
|      \hbox{an arc or a circle}\hfill\cr|\parbreak
|      \hbox{a fan or an arc}\hfill\cr}\right\}.$$|\par
\smallskip\noindent
The first example includes |\!| and |\,| to give slightly refined spacing;
but the point of the problem was to illustrate the hbox, not to fuss over
such extra details.
The last example can be done much more simply using the ideas of
Chapter~22, if you don't mind descending to the level of \TeX\ primitives;
for example, the first matrix could be replaced by ^^|\halign|
\begintt
\,\vcenter{\halign{#\hfil\cr an arc\cr a circle\cr a fan\cr}}\,
\endtt
and the second is similar.

\dangerexercise Sometimes display style is too grandiose, when the formula
being displayed is
$$y={1\over2}x$$
or something equally simple. One day B. L. ^{User} tried to remedy this by
typing it as `|$$y={\scriptstyle1\over\scriptstyle2}x$$|', but the
resulting formula
$$y={\scriptstyle1\over\scriptstyle2}x$$
wasn't at all what he had in mind. What's the right way to get ^^{one half}
^^{1/2--unslashed form} simply `$y={1\over2}x$' when you don't want big
^{fractions in displays}?
\answer |$$\textstyle y={1\over2}x$$|. \ (Switching to text style is
especially common in multiline formulas. For example, you will probably
find occasions to use ^|\textstyle| on both sides of the |&|'s within
an ^|\eqalign|.)

\dangerexercise What difference, if any, is there between the result of
typing `|$$|\<formula>|$$|' and the result of typing
`|$$\hbox{$|\<formula>|$}$$|'\thinspace?
\answer The latter formula will be in text style, not display style.
And even if you do type `|$$\hbox{$\displaystyle{|\<formula>|}$}$$|', the
results are not quite the same, as we will see later: \TeX\ will compress
the glue in `|$$|\<formula>|$$|' if the formula is too wide to fit on
a line at its natural width, but the glue inside |\hbox{...}| is frozen
at its natural width. The |\hbox| version also invokes
|\everyhbox| and |\everymath|.

\dangerexercise You may have noticed that most of the displays in this
manual are not centered; displayed material is usually aligned at the
left with the paragraph indentation, as part of the book design, because
this is an unusual book. Explain how you could typeset a formula like
$$\leftline{\indent$\displaystyle
    1-{1\over2}+{1\over3}-{1\over4}+\cdots=\ln2$}$$
^^{displays, non-centered}
that is off-center in this way.
\answer One solution is to put the formula in an hbox that occupies a full line:
\begintt
$$\leftline{\indent$\displaystyle
    1-{1\over2}+{1\over3}-{1\over4}+\cdots=\ln2$}$$
\endtt
But this takes a bit of typing. If you make the definitions
\begintt
\def\leftdisplay#1$${\leftline{\indent$\displaystyle{#1}$}$$}
\everydisplay{\leftdisplay}
\endtt
you can type `|$$|\<formula>|$$|' as usual, and the formatting will be
inserted automatically. \ (This doesn't work with equation numbers;
Appendix~D illustrates how to handle them as well.)

If you've had previous experience typing mathematical papers, you probably
have been thinking, ``What about ^{equation numbers}? When is this book
going to talk about them?'' Ah yes, now is the time to discuss those sneaky
little labels that appear off to the side of displays. If you type
\begindisplay
|$$|\<formula>|\eqno|\<formula>|$$|
\enddisplay
\TeX\ will display the first formula and it will also put an equation number
(the second formula) at the right-hand margin. For example,
\begintt
$$x^2-y^2 = (x+y)(x-y).\eqno(15)$$
\endtt
^^|\eqno|
will produce this:
$$x^2-y^2 = (x+y)(x-y).\eqno(15)$$
You can also get equation numbers at the left-hand margin, with ^|\leqno|.
For example,
\begintt
$$x^2-y^2 = (x+y)(x-y).\leqno(16)$$
\endtt
will produce this:
$$x^2-y^2 = (x+y)(x-y).\leqno(16)$$
Notice that you always give the equation number second, even when it is going
to appear at the left. Everything from the |\eqno| or |\leqno| command to
the |$$| that ends the display is the equation number. Thus, you're not
allowed to have two equation numbers in the same display; but there's a
way to get around that restriction, as we'll see later.

\danger Nowadays people are using right-hand equation numbers more and more,
because a display most often comes at the end of a sentence or clause, and
the right-hand convention keeps the number from intruding into the clause.
Furthermore, it's often possible to save space when a displayed equation
follows a short text line, since less space is needed above the display;
such savings are not possible with |\leqno|, because there's no room for
overlap. For example, there is less space above display~(15) than there is
above~(16) in our illustrations of\/ |\eqno| and |\leqno|, although the
formulas and text are otherwise identical.

\danger If you look closely at (15) and (16) above, you can see that the
displayed formulas have been centered without regard to the presence of
the equation numbers. But when a formula is large, \TeX\ makes sure that
it does not interfere with its number; the equation number may even be
placed on a line by itself.

\exercise How would you produce the following display?
$$\prod_{k\ge0}{1\over(1-q^kz)}=
    \sum_{n\ge0}z^n\bigg/\!\!\prod_{1\le k\le n}(1-q^k).\eqno(16')$$
\answer |$$\prod_{k\ge0}{1\over(1-q^kz)}=|\parbreak
        |   \sum_{n\ge0}z^n\bigg/\!\!\prod_{1\le k\le n}(1-q^k).\eqno(16')$$|

\dangerexercise Equation numbers are math formulas, typeset in text style.
So how can you get an equation number like `\hbox{(3--1)}'
(with an ^{en-dash})?
\answer |\eqno\hbox{(3--1)}|.

\ddangerexercise B. L. ^{User} tried typing `|\eqno(*)|' and `|\eqno(**)|',
and he was pleased to discover that this produced the equation numbers
`$(*)$' and `$(**)$'. \ [He had been a bit worried that they would come out
`(*)' and `(**)' instead.] \ But then a few months later he tried
`|\eqno(***)|' and got a surprise. What was it?
\answer When you type an ^{asterisk} in math mode, plain \TeX\ considers
|*| to be a binary operation. In the cases `|(*)|' and `|(**)|', the
binary operations are converted to type~Ord, because they don't appear in
a binary context; but the middle asterisk in `|(***)|' remains of type~Bin.
So the result was `$(***)$'. To avoid the extra medium spaces, you can
type `|\eqno(*{*}*)|'; or you can change ^|\mathcode||`*|, if you never use
|*| as a binary operation.

\ddanger Somewhere in this manual there ought to be a description of exactly
how \TeX\ displays formulas; i.e., how it centers them, how it places the
equation numbers, how it inserts extra space above and below, and so on.
Well, now is the time for those rules to be stated. They are somewhat
complex, because they interact with things like |\parshape|, and because
they involve several parameters that haven't been discussed yet. The purpose
of the rules is to explain exactly what sorts of boxes, glue, and
penalties are placed onto the current ^{vertical list} when a display occurs.

\ddanger If a display occurs after, say, four lines of a paragraph, \TeX's
internal register called ^|\prevgraf| will be equal to~4 when the display
starts.  The display will be assumed to take three lines, so |\prevgraf|
will become~7 when the paragraph is resumed at the end of the display
(unless you have changed |\prevgraf| in the meantime). \TeX\ assigns
special values to three \<dimen> parameters immediately after the opening
|$$| is sensed:  ^|\displaywidth| and ^|\displayindent| are set to the
line width~$z$ and the shift amount~$s$ for line number |\prevgraf|${}+2$,
based on the current paragraph shape or hanging indentation. \ (Usually
|\displaywidth| is the same as ^|\hsize|, and |\displayindent| is zero,
but the paragraph shape can vary as described in Chapter~14.) \ Furthermore,
^|\predisplaysize| is set to the effective width~$p$ of the line preceding
the display, as follows: If there was no previous line (e.g., if the |$$|
was preceded by ^|\noindent| or by the closing |$$| of another display),
$p$~is set to $-16383.99999\pt$ (i.e., to the smallest legal dimension,
$-$^|\maxdimen|).  Otherwise \TeX\ looks inside the hbox that was formed
by the previous line, and sets $p$ to the position of the right edge of
the rightmost box inside that hbox, plus the indentation by which the
enclosing hbox has been moved right, plus two ems in the current font.
However, if this value of~$p$ depends on the fact that glue in that hbox
was stretching or shrinking---for example, if the ^|\parfillskip| glue is
finite, so that the material preceding it has not been set at its natural
width---then $p$~is set to |\maxdimen|. \ (This doesn't happen
often, but it keeps \TeX\ machine independent, since $p$~never depends on
quantities that may be rounded differently on different computers.)  \
Notice that |\displaywidth| and |\displayindent| are not
affected by |\leftskip| and |\rightskip|, but
|\predisplaysize| is. The values of\/ |\displaywidth|,
|\displayindent|, and |\predisplaysize|
will be used by \TeX\ after the displayed formula has been
read, as explained below; your program can examine them and/or change
them, if you want the typesetting to be done differently.

\ddanger After a display has been read, \TeX\ converts it from a math list
to a horizontal list~$h$ in display style, as explained in
Appendix~G\null.  An equation number, if present, is processed in text
style and put into an hbox~$a$ with its natural width. Now the fussy
processing begins: Let $z$, $s$, and~$p$ be the current values of\/\
|\displaywidth|, |\displayindent|, and
|\predisplaysize|. Let $q$ and~$e$ be zero if there is no equation
number; otherwise let~$e$ be the width of the equation number, and let~$q$
be equal to $e$~plus one quad in the symbols font (i.e., in
^|\textfont||2|). Let $w_0$ be the natural width of the displayed
formula~$h$. If $w_0+q\le z$, list~$h$~is packaged in an hbox~$b$ having
its natural width~$w_0$.  But if $w_0+q>z$ (i.e., if the display is too
wide to fit at its natural width), \TeX\ performs the following ``^{squeeze
routine}'':  If $e\ne0$ and if there is enough shrinkability in the
displayed formula~$h$ to reduce its width to $z-q$, then list $h$ is
packaged in an hbox~$b$ of width~$z-q$. Otherwise $e$ is set to zero, and
list~$h$ is packaged in a (possibly overfull) hbox~$b$ of width $\min(w_0,z)$.

\ddanger (Continuation.) \ \TeX\ tries now to center the display without
regard to the equation number. But if such centering would make it too close
to that number (where ``too close'' means that the space between them is less
than the width~$e$), the equation is either centered in the remaining space
or placed as far from the equation number as possible. The latter alternative
is chosen only if the first item on list~$h$ is glue, since \TeX\ assumes that
such glue was placed there in order to control the spacing precisely.
But let's state the rules more formally: Let~$w$ be the width of box~$b$.
\TeX\ computes a displacement~$d$, to be used later when positioning box~$b$,
by first setting $d={1\over2}(z-w)$. If $e>0$ and if $d<2e$, then $d$~is
reset to ${1\over2}(z-w-e)$ or to zero, where zero is chosen if list~$h$
begins with a glue item.

\ddanger (Continuation.) \ \TeX\ is now ready to put things onto the current
vertical list, just after the material previously constructed for the
paragraph-so-far. First comes a ^{penalty} item, whose cost is an integer
parameter called ^|\predisplaypenalty|. Then comes glue. If $d+s\le p$,
or if there was a left equation number (|\leqno|), \TeX\ sets $g_a$ and
$g_b$ to glue items specified by the parameters ^|\abovedisplayskip|
and ^|\belowdisplayskip|, respectively; otherwise $g_a$ and $g_b$ become
glue items corresponding to ^|\abovedisplayshortskip| and
^|\belowdisplayshortskip|. \ [Translation: If the predisplaysize is short
enough so that it doesn't overlap the displayed formula, the glue above and
below the display will be ``short'' by comparison with the glue that is
used when there is an overlap.] \ If $e=0$ and if there is an |\leqno|,
the equation number is appended as an hbox by itself, shifted right~$s$ and
preceded by interline glue as usual; an infinite penalty is also appended,
to prevent a page break between this number and the display. Otherwise
a glue item~$g_a$ is placed on the vertical list.

\ddanger (Continuation.) \ Now comes the displayed equation itself. If
$e\ne0$, the equation number box~$a$ is combined with the formula box~$b$ as
follows: Let~$k$ be a kern of width $z-w-e-d$. In the |\eqno| case, box~$b$
is replaced by an hbox containing $(b,k,a)$; in the |\leqno| case, box~$b$
is replaced by an hbox containing $(a,k,b)$, and $d$~is set to zero. In all
cases, box~$b$ is then appended to the vertical list, shifted right by~$s+d$.

\ddanger (Continuation.) \ The final task is to append the glue or the
equation number that follows the display. If there was an |\eqno| and if
$e=0$, an infinite penalty is placed on the vertical list, followed by the
equation number box~$a$ shifted right by $s+z$ minus its width, followed
by a penalty item whose cost is the value of\/ ^|\postdisplaypenalty|.
Otherwise a penalty item for the |\postdisplaypenalty| is appended
first, followed by a glue item for~$g_b$ as specified above. \TeX\ now
adds~3 to |\prevgraf| and returns to horizontal mode, ready to resume the
paragraph.

\ddanger One consequence of these rules is that you can force an equation
number to appear on a line by itself by making its width zero, i.e.,
by saying either `|\eqno|^|\llap||{$|\<formula>|$}|' or
`|\leqno|^|\rlap||{$|\<formula>|$}|'. This makes $e=0$, and
the condition $e=0$ controls \TeX's positioning logic, as explained
in the rules just given.

\ddanger Plain \TeX\ sets |\predisplaypenalty=10000|, because fine
printers traditionally shun displayed formulas at the very top of a page.
You can change |\predisplaypenalty| and
|\postdisplaypenalty| if you want to encourage or discourage
page breaks just before or just after a display. For example,
`\hbox{|$$\postdisplaypenalty=|}\allowbreak\hbox{|-10000|\<formula>|$$|}'
will force a page break, putting the formula at the bottom line. It is better
to force a ^{page break} this way than to say ^|\eject| right after |$$...$$|;
such an eject (which follows the |\belowdisplayskip| glue below the
display) causes the page to be short, because it leaves unwanted glue at
the bottom.

\ddangerexercise Read the rules carefully and deduce the final position
of `$x=y$' in the formula
\begintt
$$\quad x=y \hskip10000pt minus 1fil \eqno(5)$$
\endtt
assuming that there is no hanging indentation. Also consider |\leqno|
instead of\/ |\eqno|.
\answer Assuming that |\hsize| is less than $10000\pt$, the natural width of
this equation will be too large to fit on a line; also, |\quad| specifies
glue at the left. Therefore `$x=y$' will appear exactly $1\em$ from
the left, and `(5)' will appear flush right. \ (The widths will satisfy
^^{displays, non-centered} $w=z-q$, $d=0$, $k=q-e=18\rm\,mu$.) \
In the case of\/ |\leqno|, `(5)' will appear flush left, followed by
one quad of space in |\textfont2|, followed by one quad of space in the
current text font, followed by `$x=y$'.

\ddanger \TeX\ also allows ``^{alignment displays},'' which are not
processed in math mode because they contain no formulas at the outer
level. An alignment display is created by commands of the general form
\begindisplay
|$$|\<assignments>|\halign{|\<alignment>|}|\<assignments>|$$|
\enddisplay
where the \<assignments> are optional things like parameter changes that do not
produce any math lists. In such displays, the |\halign| is processed exactly
as if it had appeared in vertical mode, and it will construct a vertical
list~$v$ as usual, except that each row of the alignment will be shifted
right by the ^|\displayindent|. After the alignment and the closing
assignments have been processed, \TeX\ will put a ^|\predisplaypenalty|
item and some ^|\abovedisplayskip| glue on the main vertical list, followed
by~$v$, followed by a ^|\postdisplaypenalty| item and ^|\belowdisplayskip|
glue. Thus, alignment displays are essentially like ordinary alignments,
except that they can interrupt paragraphs; furthermore, they are embedded in
glue and penalties just like other displays. The ^|\displaywidth| and
^|\predisplaysize| do not affect the result, although you could use
those parameters in your ^|\halign|. An entire alignment display is considered
to be only three lines long, as far as ^|\prevgraf| is concerned.

\subsection Multi-line displays. OK, the use of displayed formulas is
very nice. But when you try typing a lot of manuscripts you will run into
some displays that don't fit the simple pattern of a one-line formula with
or without an equation number. Plain \TeX\ provides special control
sequences that will cover most of the remaining cases.

Multi-line displays usually consist of several equations that should be
lined up by their `$=$'~signs, as in
$$\eqalign{X_1+\cdots+X_p&=m,\cr
           Y_1+\cdots+Y_q&=n.\cr}$$
The recommended procedure for such a display is to use ^|\eqalign|,
which works with special markers |&| ^^{ampersand} and ^|\cr| that we
have already encountered in connection with |\cases| and |\matrix|
in Chapter~18. Here's how to type this particular one:
\begintt
$$\eqalign{X_1+\cdots+X_p&=m,\cr
           Y_1+\cdots+Y_q&=n.\cr}$$
\endtt
There can be any number of equations in an |\eqalign|; the general pattern is
$$\halign{\indent#&#\hfil\cr
|\eqalign{|&\<left-hand side$_1$>|&|\<right-hand side$_1$>|\cr|\cr
           &\<left-hand side$_2$>|&|\<right-hand side$_2$>|\cr|\cr
\noalign{\vskip-2pt}
           &\qquad\vdots\cr
           &\<left-hand side$_n$>|&|\<right-hand side$_n$>|\cr}|\cr}$$
where each \<right-hand side> starts with the symbol on which you want
alignment to occur. For example, every right-hand side often begins
with an $=$~sign. The equations will be typeset in display style.

\exercise In practice, the left-hand sides of aligned formulas are often
blank, and the alignment is often done with respect to other symbols
as well as~$=$. For example, the following display is typical; see if you
can guess how the author typed it:
$$\eqalign{T(n)\le T(2^{\lceil\lg n\rceil})
    &\le c(3^{\lceil\lg n\rceil}-2^{\lceil\lg n\rceil})\cr
    &<3c\cdot3^{\lg n}\cr
    &=3c\,n^{\lg3}.\cr}$$ % from v2 p279
\answer (Note in particular that the final `|.|'\ comes {\sl before\/} the
final `|\cr|'.)
\begintt
$$\eqalign{T(n)\le T(2^{\lceil\lg n\rceil})
    &\le c(3^{\lceil\lg n\rceil}-2^{\lceil\lg n\rceil})\cr
    &<3c\cdot3^{\lg n}\cr
    &=3c\,n^{\lg3}.\cr}$$
\endtt

The result of\/ |\eqalign| is a vertically centered box. This makes it easy to
get a formula like
$$\left\{
\eqalign{\alpha&=f(z)\cr \beta&=f(z^2)\cr \gamma&=f(z^3)\cr}
\right\}\qquad\left\{
\eqalign{x&=\alpha^2-\beta\cr y&=2\gamma\cr}\right\}.$$ % meaningless
You simply use |\eqalign| twice in the same line:
\begintt
$$\left\{
\eqalign{\alpha&=f(z)\cr \beta&=f(z^2)\cr \gamma&=f(z^3)\cr}
\right\}\qquad\left\{
\eqalign{x&=\alpha^2-\beta\cr y&=2\gamma\cr}\right\}.$$
\endtt

\exercise Try your hand at the numbered two-line display % Polya/Szego V.29
$$\eqalign{P(x)&=a_0+a_1x+a_2x^2+\cdots+a_nx^n,\cr
   P(-x)&=a_0-a_1x+a_2x^2-\cdots+(-1)^na_nx^n.\cr}\eqno(30)$$
[{\sl Hint:\/} Use the fact that |\eqalign| produces a vertically centered
box; the equation number `(30)' is supposed to appear halfway between
the two lines.]
\answer |$$\eqalign{P(x)&=a_0+a_1x+a_2x^2+\cdots+a_nx^n,\cr|\parbreak
        |   P(-x)&=a_0-a_1x+a_2x^2-\cdots+(-1)^na_nx^n.\cr}\eqno(30)$$|\par

\exercise What happens if you forget the |&| in one equation of an |\eqalign|?
\answer Both sides of that equation are considered to be on the left, so
you get results that look like this:
$$\openup-\jot
\left\{\eqalign{\alpha&=f(z)\cr \beta&=f(z^2)\cr \gamma=f(z^3)\cr}
  \right\}.$$

\danger Multi-line formulas sometimes fit together in odd ways, and you'll
find that every once in a~while you will want to move certain lines farther
apart or closer together. If you type `^|\noalign||{|^|\vskip|\<glue>|}|'
after any |\cr|, \TeX\ will insert the given amount of extra glue just
after that particular line. For example,
\begintt
\noalign{\vskip3pt}
\endtt
will put $3\pt$ of additional space between lines. You can also change the
amount of space before the first line, in the same way.

The next level of complexity occurs when you have several aligned
equations with several equation numbers. Or perhaps some of the
lines are numbered and others are not:
$$\eqalignno{(x+y)(x-y)&=x^2-xy+yx-y^2\cr
    &=x^2-y^2;&(4)\cr
  (x+y)^2&=x^2+2xy+y^2.&(5)\cr}$$
For this situation plain \TeX\ provides ^|\eqalignno|; you use it like
|\eqalign|, but on each line that you want an equation number you add
`|&|\<equation number>' just before the |\cr|. The example above was
generated by
\begintt
$$\eqalignno{(x+y)(x-y)&=x^2-xy+yx-y^2\cr
    &=x^2-y^2;&(4)\cr
  (x+y)^2&=x^2+2xy+y^2.&(5)\cr}$$
\endtt
Notice that the second |&| is omitted unless there's an equation number.

And there's also ^|\leqalignno|, which puts equation numbers at the left.
In this case it is appropriate to move the `(4)' to the beginning
of its equation:
$$\leqalignno{(x+y)(x-y)&=x^2-xy+yx-y^2&(4)\cr
    &=x^2-y^2;\cr
  (x+y)^2&=x^2+2xy+y^2.&(5)\cr}$$
Although the equation numbers appear at the left, you are still supposed to
input them at the right, just as you do with |\leqno|; in other words,
you should type
`|$$\leqalignno{(x+y)(x-y)&...&(4)\cr...}$$|' to get the previous display.

Caution: |\eqalignno| and |\leqalignno| both center the set of equations
without regard to the widths of the equation numbers. If the equations or
their numbers get too wide, they might overlap, yet no error message will
be given.

\exercise Typeset the following display: ^^|\gcd|
$$\leqalignno{\gcd(u,v)&=\gcd(v,u);&(9)\cr
    \gcd(u,v)&=\gcd(-u,v).&(10)\cr}$$ % v2 p316
\answer |$$\leqalignno{\gcd(u,v)&=\gcd(v,u);&(9)\cr|\parbreak
        |    \gcd(u,v)&=\gcd(-u,v).&(10)\cr}$$|

\exercise And here's another one to try, just to keep in practice: ^^|\int|
$$\vbox{
\eqalignno{\biggl(\int_{-\infty}^\infty e^{-x^2}\,dx\biggr)^2
  &=\int_{-\infty}^\infty\int_{-\infty}^\infty
    e^{-(x^2+y^2)}\,dx\,dy\cr
  &=\int_0^{2\pi}\int_0^\infty e^{-r^2}r\,dr\,d\theta\cr
  &=\int_0^{2\pi}\biggl(-{e^{-r^2}\over2}
    \bigg\vert_{r=0}^{r=\infty}\,\biggr)\,d\theta\cr
  &=\pi.&(11)\cr}
}$$ % cf Joy of TeX
\answer %
|$$\eqalignno{\biggl(\int_{-\infty}^\infty e^{-x^2}\,dx\biggr)^2|\parbreak
        |  &=\int_{-\infty}^\infty\int_{-\infty}^\infty|\parbreak
        |    e^{-(x^2+y^2)}\,dx\,dy\cr|\parbreak
        |  &=\int_0^{2\pi}\int_0^\infty e^{-r^2}r\,dr\,d\theta\cr|\parbreak
        |  &=\int_0^{2\pi}\biggl(-{e^{-r^2}\over2}|\parbreak
        |    \bigg|\||_{r=0}^{r=\infty}\,\biggr)\,d\theta\cr|\parbreak
        |  &=\pi.&(11)\cr}$$| ^^|\bigg|

\danger Although |\eqalign| and |\eqalignno| look nearly the same, there's
really a fundamental distinction between them: |\eqalign| makes a single,
vertically centered box, which is no wider than it needs to be; but
|\eqalignno| generates a set of lines that have the full display width
(reaching all the way to both margins). Thus, for example, you can use
|\eqalign| several times in a display, but |\eqalignno| can appear only
once. If you try to use ^|\eqno| in conjunction with |\eqalign|,
you get a decent result, but if you try to use |\eqno| in connection
with |\eqalignno| you'll get some sort of weird error message(s).

\ddanger The definitions in Appendix~B reveal why |\eqalign| and |\eqalignno|
behave differently: |\eqalign| is an
abbreviation for ^|\vcenter||{|^|\halign||{...}}|, while
|\eqalignno| is an abbreviation for |\halign to\displaywidth{...}|;
thus the |\eqalignno| macro generates an ``^{alignment display}.''

\ddanger This difference between |\eqalign| and |\eqalignno| has two
interesting consequences. \ (1)~It's impossible to break an |\eqalign|
between pages, but an |\eqalignno| can be broken. In fact, you can
{\sl force\/} a ^{page break} after a particular line if you insert
`^|\noalign||{|^|\break||}|'
after the |\cr| for that line. You can prohibit {\sl all\/} breaks
in an |\eqalignno| if you set ^|\interdisplaylinepenalty||=10000|; or you
can enclose the whole works in a ^|\vbox|:
\begintt
$$\vbox{\eqalignno{...}}$$
\endtt
(2) You can also insert a line of text between two equations, without
losing the alignment. For example, consider the two displays
$$\eqalignno{x&=y+z\cr
  \noalign{\hbox{and}}
  x^2&=y^2+z^2.\cr}$$
These were actually generated as a single display by typing
\begintt
$$\eqalignno{x&=y+z\cr
  \noalign{\hbox{and}}
  x^2&=y^2+z^2.\cr}$$
\endtt
Therefore the fact that their $=$ signs line up is not just a lucky
coincidence. Sometimes you will want to adjust the spacing above or below
such a line of inserted text, by putting a |\vskip| or two inside of the
|\noalign{...}|.  Incidentally, this example also shows that it is
possible to use |\eqalignno| without giving any equation numbers.

\ddangerexercise What happens if\/ |\eqalign| is substituted for
|\eqalignno| in this last example?
\answer You get the displayed box
$$\eqalign{x&=y+z\cr
  \noalign{\hbox{and}}
  x^2&=y^2+z^2.\cr}$$
Reason: The `and' occurs at the left of the |\eqalign| box, not at the
left of the whole display, and the |\eqalign| box is centered as usual.

\ddangerexercise Our friend Ben ^{User} got into trouble again when he tried to
move an equation number up higher than its usual position, by typing this:
^^|\raise|
\begintt
$$\eqalignno{...&\raise6pt\hbox{(5)}\cr}$$
\endtt
What was his oversight, and what could he have done instead?
\answer By raising the equation number, he increased the line height,
so \TeX\ put extra space between that line and the previous line
when it calculated the interline glue. If he had said
`^|\smash||{\raise...}|', he wouldn't have had that problem.

\danger For other types of displays, plain \TeX\ provides ^|\displaylines|,
which lets you display any number of formulas in any way you want,
without any alignment. The general form is
$$\halign{\indent\hfil#&#\hfil\cr
|$$\displaylines{|&\<displayed formula$_1$>|\cr|\cr
    &\<displayed formula$_2$>|\cr|\cr
\noalign{\vskip-2pt}
    &\qquad\vdots\cr
    &\<displayed formula$_n$>|\cr}$$|\cr}$$
Each formula will be centered, because |\displaylines| puts ^|\hfil| at
the left and the right of each line; you can override this centering to
get things flush left or flush right by inserting ^|\hfill|, which takes
precedence over |\hfil|.

\dangerexercise Use |\displaylines| to typeset the three-line display
$$\displaylines{\hfill x\equiv x;\hfill\llap{(1)}\cr
   \hfill\hbox{if}\quad x\equiv y\quad\hbox{then}\quad
      y\equiv x;\hfill\llap{(2)}\cr
   \hfill\hbox{if}\quad x\equiv y\quad\hbox{and}\quad
      y\equiv z\quad\hbox{then}\quad
      x\equiv z.\hfill\llap{(3)}\cr}$$
\answer |$$\displaylines{\hfill x\equiv x;\hfill\llap{(1)}\cr|\parbreak
        |   \hfill\hbox{if}\quad x\equiv y\quad\hbox{then}\quad|\parbreak
        |      y\equiv x;\hfill\llap{(2)}\cr|\parbreak
        |   \hfill\hbox{if}\quad x\equiv y\quad\hbox{and}\quad|\parbreak
        |      y\equiv z\quad\hbox{then}\quad|\parbreak
        |      x\equiv z.\hfill\llap{(3)}\cr}$$|\par\medskip\noindent
There's also a trickier solution, which begins with
\begintt
$$\displaylines{x\equiv x;\hfil\llap{(1)}\hfilneg\cr
\endtt

\danger If you look closely at the multi-line displays in this chapter,
you'll see that the baselines are farther apart than they are in normal
text; mathematics publishers generally do this in order to make the
displays easier to read. In accordance with this tradition, |\eqalign| and
its relatives automatically increase the ^|\baselineskip|. If~you are
making a multi-line display with \TeX's primitive ^|\halign| command,
instead of using one of the plain \TeX\ macros, you might want to make
this same baseline adjustment, and you can do it easily by saying
`|$$\openup1\jot \halign{...}$$|'. The ^|\openup| macro increases
^|\lineskip| and ^|\lineskiplimit| as well as |\baselineskip|. If
you say `|\openup2\jot|', the lines are spread apart 2 extra units, where
plain \TeX\ opens things up in units of $3\pt$. Since |$$...$$| acts as a
^{group}, the effect of\/ |\openup| will disappear when the display is
finished. Any \<dimen> can follow |\openup|, but it's customary to express the
amount symbolically in terms of a ^|\jot| instead of using absolute units;
^^{generic coding}
then your manuscript can be used with a variety of different formats.

\ddanger Plain \TeX's ^|\displaylines|, ^|\eqalignno|, and ^|\leqalignno|
macros begin with `|\openup1\jot|'. If you don't want the lines to be
opened up, you can cancel this by saying, e.g.,
`|$$\openup-1\jot \eqalignno{...}$$|', because |\openup| has a cumulative
effect.

\ddanger Suppose that you have decided to make a homegrown display having
the general form `|$$\openup1\jot \halign{...}$$|'; and for convenience,
let's suppose that the normal conventions of plain \TeX\ are in force, so
that |\jot=3pt| and |\baselineskip=12pt|. Then the |\openup| macro
changes the baselineskip distance to $15\pt$. It follows that the baseline
of the text line that immediately precedes the display will be $15\pt$
above the topmost baseline of the display, plus the ^|\abovedisplayskip|.
But when the paragraph resumes, its next baseline will be only $12\pt$
below the bottom baseline of the display, plus the ^|\belowdisplayskip|,
because the |\baselineskip| parameter will have reverted to its normal
value. The |\eqalignno| and |\displaylines| macros say
`|\noalign{\vskip|$-d$|}|' before their first lines, where $d$ is
the net amount of opening-up, in order to compensate for this difference.

\subsection Long formulas. Our discussion of mathematics typing is almost
complete; we need to deal with just one more problem: What should be
done when a formula is so long that it doesn't fit on a single line?

For example, suppose that you encounter the equation
$$\hfuzz=20pt % overfull box tolerated here
\sigma(2^{34}-1,2^{35},1)=
  -3+(2^{34}-1)/2^{35}+2^{35}\!/(2^{34}-1)+7/2^{35}(2^{34}-1)
  -\sigma(2^{35},2^{34}-1,1).$$ % from v2, 1st ed, p76
You'll have to break it up somehow; \TeX\ has done its best to squeeze
everything together by shrinking the spaces next to the $+$ and~$-$ signs
to zero, but still the line has come out overfull.

Let's try to break that equation just before the `$+7$'. One common way to
do this is to type
\begintt
$$\eqalign{\sigma(2^{34}-1,2^{35},1)
&=-3+(2^{34}-1)/2^{35}+2^{35}\!/(2^{34}-1)\cr
&\qquad+7/2^{35}(2^{34}-1)-\sigma(2^{35},2^{34}-1,1).\cr}$$
\endtt
which yields
$$\eqalign{\sigma(2^{34}-1,2^{35},1)
  &=-3+(2^{34}-1)/2^{35}+2^{35}\!/(2^{34}-1)\cr
  &\qquad+7/2^{35}(2^{34}-1)-\sigma(2^{35},2^{34}-1,1).\cr}$$
The idea is to treat a long one-line formula as a two-line formula,
using |\qquad| on the second line so that the second part of the formula
appears well to the right of the $=$~sign on the first line.

\exercise Explain how to deal with the following display. % v2 p107
$$\eqalignno{x_nu_1+\cdots+x_{n+t-1}u_t
   &=x_nu_1+(ax_n+c)u_2+\cdots\cr
   &\qquad+\bigl(a^{t-1}x_n+c(a^{t-2}+\cdots+1)\bigr)u_t\cr
   &=(u_1+au_2+\cdots+a^{t-1}u_t)x_n+h(u_1,\ldots,u_t).
     \quad&(47)\cr}$$
\answer |$$\eqalignno{x_nu_1+\cdots+x_{n+t-1}u_t|\parbreak
        |   &=x_nu_1+(ax_n+c)u_2+\cdots\cr|\parbreak
        |   &\qquad+\bigl(a^{t-1}x_n+c(a^{t-2}+\cdots+1)\bigr)u_t\cr|\parbreak
        |   &=(u_1+au_2+\cdots+a^{t-1}u_t)x_n+h(u_1,\ldots,u_t).|\parbreak
        |     \quad&(47)\cr}$$|\par\noindent
You weren't expected to insert the `|\quad|' on the last line; such
refinements usually can't be anticipated until you see the first proofs.
But without that |\quad| the `(47)' would occur half a quad closer to the
formula.

\danger It's quite an art to decide how to ^{break long displayed formulas}
into several lines; \TeX\ never attempts to break them, because no set of
rules is really adequate. The author of a mathematical manuscript is
generally the best judge of what to do, since break positions depend on
subtle factors of mathematical exposition. For example, it is often
desirable to emphasize some of the symmetry or other structure that
underlies a formula, and such things require a solid understanding of
exactly what is going on in that formula.

\begingroup\ninepoint
\danger Nevertheless, it is possible to state a few rules of thumb about
how to deal with long formulas in displays, since there are some
principles that the best mathematical typesetters tend to follow:\enddanger

\smallskip
\textindent{a)}Although formulas within a paragraph always break {\sl after\/}
binary operations and relations, displayed formulas always break {\sl before\/}
binary operations and relations. Thus, we didn't end the first line of
our $\sigma(\,\ldots\,)$ example with `|(2^{34}-1)+|'; we ended it with
`|(2^{34}-1)|' and began the second line with `|+|'.

\smallskip
\textindent{b)}When an equation is broken before a binary operation, the second
line should start at least two quads to the right of where the innermost
subformula containing that binary operation begins on the first line.
For example, if you wish to break
\begindisplay
|$$\sum_{0<k<n}\left(|\<formula$_1$>|+|\<formula$_2$>|\right)$$|
\enddisplay
at the plus sign between \<formula$_1$> and \<formula$_2$>, it is almost
mandatory to have the plus sign on the second line appear somewhat to the
right of the large left parenthesis that corresponds to `|\left(|'.

\endgroup

\danger In the example just considered, special care is needed to break the
formula into two lines, because ^|\left| and ^|\right| delimiters cannot be
used in isolation; you can't have only |\left| in one line of a formula
and only |\right| in the second. Furthermore, you'll want the two delimiters
to be of the same size, even though they occur in different lines. The best
solution is usually to choose the delimiter size yourself; for example,
you could type
\begindisplay
|$$\eqalign{\sum_{0<k<n}\biggl(&|\<formula$_1$>|\cr|\cr
|             &\qquad{}+|\<formula$_2$>|\biggr)\cr}$$|\cr
\enddisplay
if\/ ^|\bigg| delimiters are best. Notice that the |&|~markers don't occur
at $=$~signs in this example, they just mark a point of alignment.

\danger There's another way to break long formulas, sometimes called the
{\sl^{two-line}\/} form. The idea is to put the first part of the formula
almost ^{flush left}, and to put the second part almost ^{flush right},
where ``almost flush'' means ``one quad away.'' Thus, the two-line form of
the long $\sigma(\,\ldots\,)$ equation considered earlier is
$$\displaylines{\quad\sigma(2^{34}-1,2^{35},1)
  =-3+(2^{34}-1)/2^{35}+2^{35}\!/(2^{34}-1)\hfill\cr
\hfill{}+7/2^{35}(2^{34}-1)-\sigma(2^{35},2^{34}-1,1).\quad\cr}$$
It isn't difficult to get this two-line effect with ^|\displaylines|:
\begintt
$$\displaylines{\quad\sigma(2^{34}-1,2^{35},1)
  =-3+(2^{34}-1)/2^{35}+2^{35}\!/(2^{34}-1)\hfill\cr
\hfill{}+7/2^{35}(2^{34}-1)-\sigma(2^{35},2^{34}-1,1).\quad\cr}$$
\endtt
An extra `|{}|' was typed on the second line here so that \TeX\ would know
that the `|+|' is a binary operation.  The two-line form is especially
recommended for equations that have a long left-hand side; in that case the
break generally comes just before the~$=$~sign.

\dangerexercise Typeset the following display:
$$\displaylines{\quad\sum_{1\le j\le n}{1\over
    (x_j-x_1)\ldots(x_j-x_{j-1})(x-x_j)(x_j-x_{j+1})
    \ldots(x_j-x_n)}\hfill\cr
  \hfill={1\over(x-x_1)\ldots(x-x_n)}.\quad(27)\cr}$$ % v2 p80
\answer |$$\displaylines{\quad\sum_{1\le j\le n}{1\over|\parbreak
        |    (x_j-x_1)\ldots(x_j-x_{j-1})(x-x_j)(x_j-x_{j+1})|\parbreak
        |    \ldots(x_j-x_n)}\hfill\cr|\parbreak
        |  \hfill={1\over(x-x_1)\ldots(x-x_n)}.\quad(27)\cr}$$|

\ddangerexercise If it is necessary to typeset a huge fraction like
^^{fraction, huge}
$$\def\\#1;{(#1;q^2)_\infty}
q^{{1\over2}n(n+1)}\\ea;\\eq/a;\\caq/e;\\cq^2\!/ae;
\over(e;q)_\infty(cq/e;q)_\infty$$
in a single narrow column, you might have to break up the numerator and
resort to
\begindisplay
$\displaystyle{\def\\#1;{(#1;q^2)_\infty}
\displaystyle{q^{{1\over2}n(n+1)}\\ea;\\eq/a;\qquad\atop
  \hfill\\caq/e;\\cq^2\!/ae;}
\over(e;q)_\infty(cq/e;q)_\infty}$
\enddisplay
How would you specify the latter fraction to \TeX?
% cf SIAM J Math Anal 7 (1976) p333; even longer ones appear on p334
\answer |$$\def\\#1;{(#1;q^2)_\infty} % to save typing|\parbreak
        |\displaystyle{q^{{1\over2}n(n+1)}\\ea;\\eq/a;\qquad\atop|\parbreak
        |  \hfill\\caq/e;\\cq^2\!/ae;}|\parbreak
        |\over(e;q)_\infty(cq/e;q)_\infty$$|

\endchapter

When a formula is too long for the page-width
and has to be broken into successive lines
(and we are now, of course, speaking of displayed formulae),
it should be broken, if possible, at the end of a natural `phrase';
if, for example, it is a much-bracketed formula,
it should be broken at the end of one of the major brackets
and not at an inner symbol.
This natural phrasing (as in music or speech)
makes for intelligibility between writer and reader
and should not be left to the compositor.
An author, when he finds himself writing a longish formula,
should indicate a convenient point of fracture in case of need.
\author ^{CHAUNDY}, ^{BARRETT}, and ^{BATEY}, %
 {\sl The Printing of Mathematics\/} (1954) % p38

\bigskip

Some authors use display with discretion,
some run even extremely long, complicated equations into the text,
while others tend to display every equation in the paper.
The tendency to overdisplay is probably more predominant
than the tendency to underdisplay;
for this reason it is possible for the copy editor to shorten
(and even improve) papers by running displayed material into text. $\ldots$
On the other hand, there are occasions when the copy editor needs
to suggest the display of complicated expressions that have been run into text,
particularly when it would involve a bad break at the end of a text line.
\author ELLEN ^{SWANSON}, {\sl Mathematics into Type\/} (1971) % p41

\eject
\beginchapter Chapter 20. Definitions\\(also called Macros)

You can often save time typing math formulas by letting ^{control sequences}
stand for constructions that occur frequently in a particular manuscript.
For example, if some document uses the vector `$(x_1,\ldots,x_n)$' a lot,
you can type
\begintt
\def\xvec{(x_1,\ldots,x_n)}
\endtt
and |\xvec| will henceforth be an abbreviation for `|(x_1,\ldots,x_n)|'.
Complex displays \rlap{like}
$$\def\xvec{(x_1,\ldots,x_n)}
\sum_{\xvec\ne(0,\ldots,0)} \bigl(f\xvec+g\xvec\bigr)$$
can then be typed simply as
\begintt
$$\sum_{\xvec\ne(0,\ldots,0)} \bigl(f\xvec+g\xvec\bigr)$$
\endtt
instead of in a tedious long form. By ^{defining a control sequence}
like |\xvec|, you not only cut down on the number of keystrokes that you need
to make, you also reduce your chances of introducing typographical errors
and inconsistencies. ^^{abbreviations, see macros}

Of course, you usually won't be making a definition just to speed up the
typing of one isolated formula; that doesn't gain anything, because time goes by
when you're deciding whether or not to make a definition, and when you're
typing the definition itself. The real payoff comes when some cluster of
symbols is used dozens of times throughout a manuscript. A wise typist will
look through a document before typing anything, thereby getting a feeling
for what sorts of problems will arise and what sorts of definitions will
be helpful. For example, Chapter~16 recommends that the control sequence
|\Ahat| be defined at the beginning of any manuscript that makes frequent
use of the symbol~$\hat A$.

Abbreviations like |\xvec| turn out to be useful in many applications of
computers, and they have come to be known as {\sl^{macros}\/} because they
are so powerful; one little macro can represent an enormous amount of
material, so it has a sort of macroscopic effect. System programs like
\TeX\ that are designed to deal with macro definitions are said to {\sl
expand\/} the user's macros; for example, |\xvec| expands into
|(x_1,\ldots,x_n)|, and ^|\ldots| in turn is a macro that expands into
|\mathinner{\ldotp\ldotp\ldotp}|.  Thus, |\xvec| is actually an
abbreviation for `|(x_1,\mathinner{\ldotp\ldotp\ldotp},x_n)|'.  \ (The
expansion stops here, because ^|\mathinner| is a primitive control
sequence of \TeX, and because |\ldotp| has been defined with
^|\mathchardef|; thus |\mathinner| and |\ldotp| are not macros.)

\TeX\ users generally build up their own personal ^{library of macros} for
things that they want to do in different documents. For example, it is common
to have a file called |macros.tex| that contains definitions of your favorite
special control sequences, perhaps together with commands that load your
favorite special fonts, etc. If you begin a document with the command
\begintt
\input macros
\endtt
then \TeX\ will read all those definitions, ^^|\input|
saving you all the trouble of retyping them. Of course, \TeX's memory is
limited, and it takes time to read a file, so you shouldn't put thousands
of definitions into |macros.tex|. A large collection of macro definitions
(e.g., the set of definitions in Appendix~B) is called a {\sl^{format}\/}
(e.g., ``plain \TeX\ format''); \TeX\ has a special way to input
a format at high speed, assuming that the format doesn't change very often.

The |\xvec| and |\Ahat| examples apply to math formulas, but you can make
good use of macro definitions even when you aren't doing any math at all.
For example, if you are using \TeX\ for ^{business correspondence},
you can have a |\yours| macro that stands for `Sincerely yours, A.~U.
^{Thor}'. If you often write ^{form letters} you can have macros that
generate entire sentences or paragraphs or groups of paragraphs.
The ^{Internal Revenue Service} could, for example, make use of the
following two macros:
\begintt
\def\badcheck{A penalty has been added because your
  check to us was not honored by your bank.\par}
\def\cheater{A penalty of 50\% of the underpaid tax
  has been added for fraud.\par}
\endtt
Simple macro definitions, like these, start with `|\def|'; then comes the
control sequence name, e.g., `|\badcheck|'; and then comes the replacement
text enclosed in `|{|' and~`|}|'. The ^{braces} do not represent
^{grouping} in this case; they simply show the extent of the replacement
text in the definition.  You could, of course, define a macro that
includes actual braces in its replacement text, as long as those braces
match each other properly. For example, `|\def\xbold{{\bf x}}|'
makes |\xbold| an abbreviation for `|{\bf x}|'.

\exercise
Write a |\punishment| macro that prints 100 lines containing the message
`I must not talk in class.' \ [{\sl Hint:\/} First write a macro |\mustnt|
that prints the message once; then write a macro |\five| that prints it
five times.] \checkequals\punishexno\exno
\answer |\def\mustnt{I must not talk in class.\par}|\parbreak
        |\def\five{\mustnt\mustnt\mustnt\mustnt\mustnt}|\parbreak
        |\def\twenty{\five\five\five\five}|\parbreak
        |\def\punishment{\twenty\twenty\twenty\twenty\twenty}|\par
\smallskip\noindent Solutions to more complicated problems of this type are
discussed later.

\dangerexercise What is the expansion of\/ |\puzzle|, given the following
definitions?
\begintt
\def\a{\b}
\def\b{A\def\a{B\def\a{C\def\a{\b}}}}
\def\puzzle{\a\a\a\a\a}
\endtt
\answer |ABCAB|. \ (The first |\a| expands into |A\def\a{B...}|; this redefines
|\a|, so the second |\a| expands into |B...|, etc.) \ At least, that's what
happens if\/ |\puzzle| is encountered when \TeX\ is building a list. But if
|\puzzle| is expanded in an ^|\edef| or ^|\message| or something like that,
we will see later that the interior |\def| commands are not performed
while the expansion is taking place, and the control sequences following
|\def| are expanded; so the result is an infinite string
\begintt
A\def A\def A\def A\def A\def A\def A\def A\def A...
\endtt
which causes \TeX\ to abort because the program's input stack is finite.
This example points out that a control sequence (e.g., |\b|) need not be
defined when it appears in the replacement text of a definition. The example
also shows that \TeX\ doesn't expand a macro until it needs to.

\danger As soon as you get the hang of simple macros like those illustrated
above, you will probably begin to think, ``Boy, wouldn't it be nice if I could
have a macro in which some of the text in the expansion is changeable? I'd like
to be able to stick different things into the middle of that text.'' Well,
\TeX\ has good news for you: Control sequences can be defined in terms of
{\sl^{parameters}}, and you can supply {\sl^{arguments}\/} that will be
substituted for the parameters.

\danger For example, let's consider |\xvec| again. Suppose that you not only
refer to `$(x_1,\ldots,x_n)$', but you also make frequent use of
`$(y_1,\ldots,y_n)$' and other similar things. Then you might want to type
\begintt
\def\row#1{(#1_1,\ldots,#1_n)}
\endtt
\def\row#1{(#1_1,\ldots,#1_n)}%
after which |\row x| will produce `$\row x$' and |\row y| will produce
`$\row y$'. The symbol |#1| ^^{sharpsign} stands for the first parameter to
the macro, and when you say `|\row|~|x|' the |x| is a so-called argument that
will be inserted in place of the |#1|'s in the replacement text. In this
case the argument consists of a single letter, |x|. You can also say
|\row\alpha|, in which case the argument will be the control sequence
^|\alpha|, and the result will be `$\row\alpha$'. If you want the argument
to contain more than one symbol or control sequence, you can simply enclose
it in ^{braces}; for example, |\row{x'}| yields $\row{x'}$. The argument
in this case is |x'| (without the braces). Incidentally, if you say
|\row{{x'}}|, you get $\row{{x'}}$; the reason is that only one pair of
braces is stripped off when the argument is collected, and
$({x'}_1,\ldots,{x'}_n)$ is what you get from
|({x'}_1,\ldots,{x'}_n)| in math mode, according to the rules
of Chapter~16. ^^{apostrophe}

\dangerexercise Continuing this example, what is the result of
|$\row{\bf x}$|?
\answer \def\row#1{(#1_1,\ldots,#1_n)}$\row{\bf x}$. Note that the
subscripts are bold here, because the expansion |(\bf x_1,\ldots,\bf x_n)|
doesn't ``turn off'' ^|\bf|. To prevent this, one should write
|\row{{\bf x}}|; or (better), |\row\xbold|, in conjunction with
|\def\xbold{{\bf x}}|.

\danger The notation `|#1|' suggests that there might be an opportunity to have
more than one parameter, and indeed there is. You can write, for example,
\begintt
\def\row#1#2{(#1_1,\ldots,#1_#2)}
\endtt
\def\row#1#2{(#1_1,\ldots,#1_#2)}%
after which `|\row xn|' would be the proper protocol for `$\row xn$'. There can
be as many as nine parameters, |#1| to~|#9|, and when you use them you must
number them~in order. For example, you can't use |#5| in a definition unless
the previous parameter in~that definition was called |#4|. \ (This restriction
applies only to the initial statement of parameters, before the replacement
text starts; the stated parameters can be used any number of times, in any
order, in the replacement text itself.)

\danger A control sequence has only one definition at a
time, so the second definition of\/ |\row| would supersede the first one if
both had appeared in the same document. Whenever \TeX\ encounters a macro
that it wants to expand, it uses the most recent definition. However,
definitions are ^{local} to the group that contains them; old definitions
will be restored in the usual way when a ^{group} ends.

\danger Caution: When you define a macro with simple parameters, as in
these examples, you must be careful not to put blank spaces before the
`|{|' that begins the replacement text. For example, `|\def\row #1 #2 {...}|'
will not give the same result as `|\def\row#1#2{...}|', because the spaces
after |#1| and~|#2| tell \TeX\ to look for arguments that are followed by
spaces. \ (Arguments can be ``delimited'' in a fairly general way, as
explained below.) \ But the space after |\row| is optional, as usual,
because \TeX\ always disregards spaces after control words. After
you have said `|\def\row#1#2{...}|', you are allowed to put spaces
between the arguments (e.g., `|\row x n|'), because \TeX\ doesn't
use single spaces as undelimited arguments.

\danger The following exercise is particularly recommended for people who
want to learn to write \TeX\ macros. Even if you have gotten into the
dangerous habit of skimming other exercises, you should try your hand
at this one.

\dangerexercise Extending exercise 20.\punishexno, write a ``generalized
punishment'' macro that has two parameters, so that |\punishment{run}{the
halls}| will produce 100 paragraphs that say `I~must not run in the halls.'
\answer The catch is that the parameters have to percolate down to the
|\mustnt| macro, if you extend the previous answer:
\begintt
\def\mustnt#1#2{I must not #1 in #2.\par}
\def\five#1#2{\mustnt{#1}{#2}...\mustnt{#1}{#2}}
\def\twenty#1#2{\five{#1}{#2}...\five{#1}{#2}}
\def\punishment#1#2{\twenty{#1}{#2}...\twenty{#1}{#2}}
\endtt
When you pass parameters from one macro to another in this way, you need to
enclose them in braces as shown. But actually this particular solution
punishes \TeX\ much more than it needs to, because it takes a lot of
time to copy the parameters and read them again and again. There's a
much more efficient way to do the job, by defining control sequences:
\begintt
\def\mustnt{I must not \doit\ in \thatplace.\par}
\def\punishment#1#2{\def\doit{#1}\def\thatplace{#2}%
  \twenty\twenty\twenty\twenty\twenty}
\endtt
and by defining |\five| and |\twenty| without parameters as before.
You can also delve more deeply into \TeX nicalities, constructing solutions
that are more efficient yet; \TeX\ works even faster when macros
communicate with each other via ^{boxes}.
^^{efficient macros} ^^{communication between macros}
For example,
\begintt
\def\mustnt{\copy0 }
\def\punishment#1#2{\setbox0=
  \vbox{\strut I must not #1 in #2.\strut}%
  \twenty\twenty\twenty\twenty\twenty}
\endtt
sets 100 identical paragraphs at high speed, because \TeX\ has to
process the paragraph and break it into lines only once. It's much faster
to ^{copy a box} than to build it up from scratch. \ (The ^{struts} in
this example keep the interbaseline distances correct between boxed
paragraphs, as explained in Chapter~12. Two struts are used, for if the
message takes more than one line there will be a strut at both top
and bottom. If it were known that each sentence will occupy only a single
line, no struts would be needed, because interline glue is added as
usual when a box created by |\copy| is appended to the current vertical list.)

\ninepoint % the rest of this chapter is all dangerous
\ddanger \TeX\ also allows you to define macros whose parameters are delimited
in quite a general way; you needn't always enclose arguments in braces.
For example,
\begintt
\def\cs #1. #2\par{...}
\endtt
defines a control sequence |\cs| with two parameters, and its two arguments
will be determined as follows: |#1| will consist of all tokens between
|\cs| and the next subsequent appearance of `|.|\]' (period and space);
|#2| will consist of all tokens between that `|.|\]' and the next
|\par| token. \ (The ^|\par| might be given explicitly, or it might be
generated by a blank line as explained in Chapter~8.) \ For example, when
\TeX\ expands
\begintt
\cs You owe \$5.00. Pay it.\par
\endtt
the first argument is `|You owe \$5.00|' and the second is `|Pay it.|'.
The period in `|\$5.00|' doesn't stop |#1|, in this example,
because \TeX\ keeps going until finding a period that is followed immediately
by a space.

\ddanger Furthermore, an argument will not stop when its delimiter is enclosed
in braces, because that would produce unbalanced braces. For example, in
\begintt
\def\cs #1.#2\par{...}
\endtt
the first argument is now delimited by a single period, so |#1| would be
`|You owe \$5|' and the |#2| would be `|00. Pay it.|' if\/ |\cs| were
invoked as above. But
\begintt
\cs You owe {\$5.00}. Pay it.\par
\endtt
satisfactorily hides the first period, making it part of argument |#1|,
which becomes \hbox{`|You owe {\$5.00}|'}.

\ddanger If you are designing a format for mathematical papers, you will
probably want to include a macro for the statement of ^{theorems},
definitions, lemmas, corollaries, and such things. For example, you might
want to typeset a statement like\enddanger

\proclaim Theorem 1. \TeX\ has a powerful macro capability.\par
\noindent from the input
\begintt
\proclaim Theorem 1. \TeX\ has a powerful macro capability.\par
\endtt
In fact, plain \TeX\ includes a ^|\proclaim| macro that does just that;
its definition is
\begintt
\def\proclaim #1. #2\par{\medbreak
  \noindent{\bf#1.\enspace}{\sl#2}\par\medbreak}
\endtt
^^{enunciations, see proclaim} ^^{enspace}
so the arguments are delimited exactly as in our first |\cs| example. The
replacement text here uses |\medbreak| to separate the proclaimed
paragraph from what precedes and follows; the title of the proclamation is
set in bold face type, while the text itself is set slanted.  \ (The
actual definition of\/ |\proclaim| in Appendix~B is not quite the same as
this; the final |\medbreak| has been modified so that a break between
pages will be discouraged immediately following the statement of a
theorem. Hence a short theorem will tend to appear at the top of a page
rather than at the bottom.)

\ddanger By making changes to the |\proclaim| macro, you can change the
format of all the proclamations in your paper, without changing the text
of the paper itself. For example, you could produce something like\enddanger

\medbreak
\font\tencsc=cmcsc10
\noindent {\tencsc Theorem 1:}\enspace
{\it\TeX\ has a powerful macro capability.}
\goodbreak\medbreak\noindent
by making simple alterations to the replacement text of\/ |\proclaim|,
assuming that you have a ``^{caps and small caps}'' font. \TeX\ is
intended to support ^{higher-level languages for composition} in which all of
the control sequences that a user actually types are macros rather than
\TeX\ primitives. The ideal is to be able to describe important classes of
documents in terms of their components, without mentioning actual fonts or
point sizes or details of spacing; a single ^{style-independent document}
^^{format-independent document}^^{generic coding}
can then be set in many different styles.

\ddanger Now that we have seen a number of examples, let's look at the
precise rules that govern \TeX\ macros. Definitions have the general form
\begindisplay
|\def|\<control sequence>\<parameter text>|{|\<replacement text>|}|
\enddisplay
where the \<parameter text> contains no ^{braces}, and where all occurrences
of |{| and |}| in the \<replacement text> are properly nested. Furthermore
the |#| symbol has a special significance: In the \<parameter text>, the
first appearance of |#| must be followed by~|1|, the next by~|2|, and
so on; up to nine |#|'s are allowed. In the \<replacement text> each~|#|
must be followed by a digit that appeared after~|#| in the \<parameter text>,
or else the~|#| should be followed by another~|#|. The latter case stands
for a single~|#| token when the macro is expanded; the former case stands
for insertion of the corresponding argument.

\ddanger For example, let's consider a ``random'' definition that doesn't
do anything useful except that it does exhibit \TeX's rules. The definition
\begintt
\def\cs AB#1#2C$#3\$ {#3{ab#1}#1 c##\x #2}
\endtt
says that the control sequence |\cs| is to have a parameter text consisting of
nine tokens
\begindisplay
|A|$_{11}$, \ |B|$_{11}$, \ |#1|, \ |#2|, \ |C|$_{11}$, \
  |$|$_3$, \ |#3|, \ \cstok{\char`$}, \ \]$_{10}$
\enddisplay
(assuming the ^{category codes} of plain \TeX), and a replacement text
of twelve tokens
\begindisplay
|#3|, \ |{|$_1$, \ |a|$_{11}$, \ |b|$_{11}$, \ |#1|, \
  |}|$_2$, \ |#1|, \ \]$_{10}$, \ |c|$_{11}$, \ |#|$_6$, \ \cstok{x}, \ |#2|.
\enddisplay
Henceforth when \TeX\ reads the control sequence |\cs| it will expect that
the next two tokens will be |A|$_{11}$ and |B|$_{11}$ (otherwise you will
get the error message `|Use| |of| |\cs| |doesn't| |match| |its|
|definition|'); then comes argument~|#1|, followed by argument~|#2|,
then~|C|$_{11}$, then~|$|$_3$, then argument~|#3|, then |\$|, and finally
a space token. It is customary to use the word ``argument'' to mean the
string of tokens that gets substituted for a parameter; parameters appear
in a definition, and arguments appear when that definition is used. \ (For
the purposes of these rules, we are extending Chapter~7's definition of
^{token}: In addition to control sequences and (character code, category
code) pairs, \TeX\ also recognizes ``^{parameter tokens},'' denoted here
by |#1|~to~|#9|. Parameter tokens can appear only in token lists for macros.)

\ddanger How does \TeX\ determine where an argument stops, you ask.
Answer: There are two cases. A {\sl^{delimited parameter}\/} is followed
in the \<parameter text> by one or more non-parameter tokens, before
reaching the end of the parameter text or the next parameter token;
in this case the corresponding argument is the shortest (possibly empty)
sequence of tokens with properly nested |{...}| groups that is followed in
the input by this particular list of non-parameter tokens. \ (Category
codes and character codes must both match, and control sequence names
must be the same.) \ An {\sl^{undelimited parameter}\/} is followed immediately
in the \<parameter text> by a parameter token, or it occurs at the very end
of the parameter text; in this case the corresponding argument is the
next nonblank token, unless that token is `|{|', when the argument will
be the entire |{...}| group that follows. In both cases, if the argument
found in this way has the form `|{|\<nested tokens>|}|', where \<nested
tokens> stands for any sequence of tokens that is properly nested with
respect to braces, the outermost braces enclosing the argument are removed
and the \<nested tokens> will remain. For example, let's continue with
|\cs| as defined above and suppose that the subsequent text contains
\begintt
\cs AB {\Look}C${And\$ }{look}\$ 5.
\endtt
Argument |#1| will be the token \cstok{Look}, since |#1| is an undelimited
parameter (it is followed immediately by~|#2| in the definition);
in this case \TeX\ ignores the blank space after |B|, and strips the
braces off~of |{\Look}|. Argument~|#2| will be empty, since |C$| follows
immediately. And argument~|#3| will be the thirteen tokens corresponding
to the text |{And\$|\]|}{look}|, because |#3| is to be followed by `|\$|\]',
and because the first occurrence of `|\$|\]' is within braces. Even though
argument~|#3| begins with a left brace and ends with a right brace, the
braces are not removed, since that would leave the unnested tokens `|And\$
}{look|'.  The net effect then, after substituting arguments for
parameters in the replacement text, will be that \TeX\ will next read the
token list
\begintt
{And\$ }{look}{ab\Look}\Look|]c#\x5.
\endtt
The space \] here will be part of the resulting token list, even though
it follows the control word |\Look|, because ^{spaces} are removed
after ^{control word} tokens only when \TeX\ first converts input lines to
token lists as described in Chapter~8.

\ddangerexercise The example definition of\/ |\cs| includes a |##| in
its replacement text, but the way |##| is actually used in that example
is rather pointless. Give an example of a definition where |##| serves
a useful purpose. ^^{sharp sharp}
\answer The |##| feature is indispensable when the replacement text of
a definition contains other definitions. For example, consider
\begintt
\def\a#1{\def\b##1{##1#1}}
\endtt
after which `|\a!|'\ will expand to `|\def\b#1{#1!}|'. We will see later
that |##| is also important for alignments; see, for example, the definition
of\/ |\matrix| in Appendix~B.

\ddanger A special extension is allowed to these rules: If the very
last character of the \<parameter text> is~|#|, so that this~|#| is
immediately followed by~|{|, \TeX\ will behave as if the~|{| had been
inserted at the right end of both the parameter text and the replacement text.
For example, if you say `|\def\a#1#{\hbox to #1}|', the
subsequent text `|\a3pt{x}|' will expand to `|\hbox to 3pt{x}|', because
the argument of\/ |\a| is delimited by a left brace.
^^{dimensions as arguments}

\ddanger Tokens that precede the first parameter token in the \<parameter
text> of a definition are required to follow the control sequence; in
effect, they become part of the control sequence name. For example, the author
might have said
\begintt
\def\TeX/{...}
\endtt
instead of defining ^|\TeX| without the slash. Then it would be necessary
to type |\TeX/| each time the \TeX\ logo is desired, but the new definition
would have the advantage that spaces are {\sl not\/} ignored after
`|\TeX/|'. You can use this idea to define macros that are intended to
be used in sentences, so that users don't have to worry about the possible
disappearance of ^{spaces}.

\ddangerexercise Define a control sequence |\a| such that |\a{...}| expands
to |\b{...}|, and such that \TeX\ gives an error message if\/ |\a| is not
immediately followed by a left brace.
\answer |\def\a#{\b}|.

\ddanger Complicated macros have a habit of behaving differently from what
you expect, when you first define them, even though \TeX's rules are
not especially complicated. If you have trouble understanding why some
|\def| doesn't work the way you think it should, help is available:
You can set |\tracingmacros=1|, whereupon \TeX\ will write something in
your log file whenever it expands a macro, and whenever it has read a
macro argument. For example, if\/ ^|\tracingmacros| is positive when
\TeX\ processes the |\cs| example above, it will put the following
four lines into the log: ^^{debugging macros}
\begintt
\cs AB#1#2C$#3\$ ->#3{ab#1}#1 c##\x #2
#1<-\Look
#2<-
#3<-{And\$ }{look}
\endtt

\ddanger In all of the rules stated above, `|{|' and `|}|' and `|#|' stand
for any characters whose ^{category codes} are respectively 1, 2, and 6 in
the token list when \TeX\ reads the macro definition; there's nothing
sacred about the particular symbols that plain \TeX\ uses to denote
grouping and parameters.  You can even make use of several different
characters with these category codes, all at the same time.

\ddangerexercise Suppose that `|[|', `|]|', and `|!|'\ have the respective
catcodes 1,~2, and~6, as do `|{|',~`|}|', and~`|#|'. See if you can guess
what the following definition means:
\begintt
\def\!!1#2![{!#]#!!2}
\endtt
What token list will result when `|\! x{[y]][z}|' is expanded?
\answer Let's go slowly on this one, so that the answer will give enough
background to answer all similar questions. The \<parameter text> of the
definition consists of the three tokens |!1|, |#2|, |[|$_1$; the
\<replacement text> consists of the six tokens |{|$_1$, |#|$_6$, |]|$_2$,
|!|$_6$, |#2|, |[|$_1$. \ (When two tokens of category~6 occur in the
replacement text, the character code of the second one survives; the
character code of a category-6 character is otherwise irrelevant. Thus,
`|\def\!!1#2#[{##]!!#2]|' would produce an essentially identical
definition.) \ When expanding the given token list, argument~|#1| is
|x|$_{11}$, since it is undelimited. Argument~|#2| is delimited by~|[|$_1$,
which is different from~|{|$_1$, so it is set provisionally to |{[y]]|;
but the outer ``braces'' are stripped off, so |#2|~reduces to the
three tokens |[|$_1$, |y|$_{11}$,~|]|$_2$. The result of the expansion
is therefore
\begindisplay
|{|$_1$ |#|$_6$ |]|$_2$ |!|$_6$ |[|$_1$ |y|$_{11}$ |]|$_2$
  |[|$_1$ |z|$_{11}$ |}|$_2$.
\enddisplay
Incidentally, if you display this with ^|\tracingmacros||=1|, \TeX\ says
\begintt
\!!1#2[->{##]!!#2[
!1<-x
#2<-[y]
\endtt
Category codes are not shown, but a character of category~6 always
appears twice in succession. A parameter token in the replacement text
uses the character code of the final parameter in the parameter text;
^^{token lists, as displayed by TeX}
`|!1|' would have been rendered `|#1|'.

\ddanger In practice, we all make mistakes. And one of the most common
typographic errors is to forget a~`|}|', or to insert an extra~`|{|',
somewhere in an argument to a macro. If \TeX\ were to follow the rules
blindly in such a case, it would have to keep absorbing more and
more tokens in hopes of finding the end of the argument. But a mistyped
argument is unending, like so many arguments in real life (sigh); so
\TeX\ would have to go on until the end of the file, or (more likely)
until tokens completely fill the computer's memory. In either case, a
single typographical error would have ruined the run, and the user would
be forced to start over. Therefore \TeX\ has another rule, intended
to confine such errors to the paragraph in which they occur: {\sl The
token `\thinspace^|\par|' is not allowed to occur as part of an argument},
unless you explicitly tell \TeX\ that |\par| is OK. Whenever \TeX\ is
about to include |\par| as part of an argument, it will abort the current
macro expansion and report that a ``^{runaway} argument'' has been found.

\ddanger If you actually want a control sequence to allow arguments with
|\par| tokens, you can define it to be a ``long'' macro by saying
`^|\long|' just before `|\def|'. For example, the |\bold| macro
defined by
\begintt
\long\def\bold#1{{\bf#1}}
\endtt
is capable of setting several paragraphs in boldface type. \ (However,
such a macro is not an especially good way to typeset bold text.
It would be better to say, e.g.,
\begintt
\def\beginbold{\begingroup\bf}
\def\endbold{\endgroup}
\endtt
because this doesn't fill \TeX's memory with a long argument.)

\ddanger The |\par|-forbidding mechanism doesn't catch all conceivable
missing-brace errors, however; you might forget the |}| at the end of a
|\def|, and the same problem would arise. In this case it's harder to
confine the error, because |\par| is a useful thing in replacement
texts; we wouldn't want to forbid |\par| there, so \TeX\ has another
mechanism:  When a macro definition is preceded by `^|\outer|', the
corresponding control sequence will not be allowed to appear in any place
where tokens are being absorbed at high speed. ^^{forbidden control sequence}
An |\outer| macro cannot appear in an argument (not even when |\par| is
allowed), nor can it appear in the parameter text or the replacement text
of a definition, nor in the ^{preamble} to an alignment, nor in ^{conditional
text} that is being skipped over.  If an |\outer| macro does show up in
such places, \TeX\ stops what it is doing and reports either a
``runaway'' situation or an ``^{incomplete}'' conditional. The ^{end of an
input file} or alignment template ^^{endtemplate}
is also considered to be |\outer| in this sense; for example, a
file shouldn't end in the middle of a definition. If you are designing a
format for others to use, you can help them detect errors before too much
harm is done, by using |\outer| with all control sequences that should
appear only at ``quiet times'' within a document. For example, Appendix~B
defines ^|\proclaim| to be |\outer|, since a user shouldn't be stating a
theorem as part of a definition or argument or preamble.

\ddanger We have now seen that |\def| can be preceded by |\long|
or~|\outer|, and it can also be preceded by ^|\global| if the definition
is supposed to transcend its group. These three prefixes can be applied
to |\def| in any order, and they can even appear more than once.
\TeX\ also has a ^|\gdef| primitive that is equivalent to |\global\def|.
Thus, for example,
\begintt
\long\outer\global\long\def
\endtt
means the same thing as `|\outer\long\gdef|'.

\ddanger So far in this manual we have encountered several ways to
assign a meaning to a control sequence. For example,
$$\halign{\indent#\hfil\quad&#\hfil\cr
|\font\cs=|\<external font name>&makes |\cs| a font identifier;\cr
|\chardef\cs=|\<number>&makes |\cs| a character code;\cr
|\countdef\cs=|\<number>&makes |\cs| a |\count| register;\cr
|\def\cs...{...}|&makes |\cs| a macro.\cr
\noalign{\medskip \hbox{It's time now to reveal another important command
  of this type:} \medskip}
|\let\cs=|\<token>&gives |\cs| the token's current meaning.\cr}$$
^^|\let|
If the \<token> is another control sequence, |\cs| will acquire the
same significance as that control sequence. For example, if you say
`|\let\a=\def|', you could then say `|\a\b...{...}|' to define a macro~|\b|,
because |\a| would behave like \TeX's primitive |\def| command.
If you say
\begintt
\let\a=\b \let\b=\c \let\c=\a
\endtt
you have interchanged the former meanings of\/ |\b| and |\c|. And if you say
\begintt
\outer\def\a#1.{#1:}
\let\b=\a
\endtt
the effect is exactly the same as `|\outer\def\b#1.{#1:} \let\a=\b|'.

\ddanger If the \<token> in a |\let| is a single character---i.e.,
if it is a (character code, category code) pair---then
the control sequence will behave to a certain extent like that
character; but there are some differences. For example, after
`|\let\zero=0|' you can't use |\zero| in a numerical constant, because
\TeX\ requires the tokens in a numerical constant to be digits, after
macro expansion; |\zero| is not a macro, so it doesn't expand. However,
such uses of\/ |\let| have their value, as we will see later.

\ddangerexercise Is there a significant difference between `|\let\a=\b|'
and `|\def\a{\b}|'?
\answer Yes indeed. In the first case, |\a| receives the meaning of\/~|\b|
that is current at the time of the |\let|. In the second case, |\a|~becomes
a~macro that will expand into the token~|\b| whenever |\a|~is used,
so it has the meaning of\/~|\b| that is current at the time of use.
You need |\let|, if you want to interchange the meanings of\/ |\a| and~|\b|.

\ddangerexercise Experiment with \TeX\ to discover the answers to the
following questions: (a)~If the control sequence ^|\par| has been redefined
(e.g., `|\def\par{\endgroup\par}|'\thinspace), is |\par| still forbidden
to appear in an argument? \ (b)~If you say |\let\xpar=\par|, is |\xpar|
also forbidden in an argument?
\answer (a) Yes. \ (b) No; any other control sequence can appear
(except those declared as |\outer| macros).

\ddanger \TeX\ also allows the construction
`^|\futurelet||\cs|\<token$_1$>\<token$_2$>', which has the effect of
`|\let\cs = |\<token$_2$>\<token$_1$>\<token$_2$>'. The idea is that you can
say, for example, `|\futurelet\a\b|' at the end of the replacement text of
a macro; \TeX~will set |\a| to the token that follows the macro, after
which |\b| will be expanded. The control sequence~|\b| can continue the
processing, and it can examine |\a| to see what's coming up next.
^^{looking ahead}

\danger The next thing a person wants, after getting used to macros with
parameters, is the ability to write macros that change their behavior
depending on current conditions. \TeX\ provides a variety of primitive
commands for this purpose. The general form of such ``^{conditional text}'' is
\begindisplay
|\if|\<condition>\<true text>|\else|\<false text>|\fi|
\enddisplay
where the \<true text> is skipped unless the \<condition> is true, and
the \<false text> is skipped unless the \<condition> is false.
If the \<false text> is empty, you can omit the~^|\else|. The
`|\if|\<condition>' part of this construction begins with a control
sequence whose first two letters are `|if|'; for example,
\begintt
\ifodd\count0 \rightpage \else\leftpage \fi
\endtt
^^|\ifodd|
specifies a condition that is true when \TeX's integer register ^|\count||0|
is odd. Since \TeX\ generally keeps the current ^{page number} in
|\count0|, the macro |\rightpage| will be expanded in this example if
the page number is odd, while |\leftpage| will be expanded if the
page number is even. Conditional commands always end with a final `^|\fi|'.

\danger Conditionals are primarily intended for experienced \TeX\ users,
who want to define high-level macros; therefore
the remaining paragraphs in this chapter are headed by ``double
dangerous bends.'' Do not feel guilty about skipping right to Chapter~21;
in other words, imagine that the manual says `|\ifexperienced|' right
here, and that there is a matching `|\fi|' at the end of the present chapter.

\ddanger Before we discuss \TeX's repertoire of\/ |\if...|\ commands, let's
look at another example, so that the general ideas will be clear. Suppose
that the |\count| register |\balance| holds an amount that somebody
has paid in excess of his or her income tax; this amount is given in
pennies, and it might be positive, negative, or zero. Our immediate goal
will be to write a \TeX\ macro that generates a suitable statement for the
^{Internal Revenue Service} to include as part of a letter to that person,
based on the amount of the balance. The statement will be quite different
for positive balances than for negative ones, so we can exploit \TeX's ability
to act conditionally:
\begintt
\def\statement{\ifnum\balance=0 \fullypaid
  \else\ifnum\balance>0 \overpaid
       \else\underpaid
       \fi
  \fi}
\endtt
Here ^|\ifnum| is a conditional command that compares two numbers; the
|\statement| macro reduces to |\fullypaid| if the balance is zero, and so on.

\ddanger It is vastly important to notice the spaces after the |0|'s in
this construction. If the example had said
\begintt
...=0\fullypaid...
\endtt
then \TeX\ would have begun to expand `|\fullypaid|' before it knew the
value of the constant |0|, because |\fullypaid| might start with a~|1| or
something that would change the number. \ (After all, `|01|' is a perfectly
acceptable \<number>, in \TeX's eyes.) \ In this particular case the
program would still have worked, because we will see in a moment that
|\fullypaid| begins with the letter~|Y|; thus, the only problem caused by
the missing space would be that \TeX\ would go slower, since it would have
to skip over the whole expansion of\/ |\fullypaid| instead of just skipping
|\fullypaid| as a single, unexpanded token. But in other situations a
missing space like this might cause \TeX\ to expand macros when you don't
want any expansion, and such anomalies can cause subtle and confusing errors.
For best results, {\sl always put a blank space after a numeric constant\/};
this blank space tells \TeX\ that the constant is complete, and
such a space will never ``get through'' to the output.  In fact, when you
don't have a blank ^{space after a constant}, \TeX\ actually has to do
more work, because each constant continues until a non-digit has been
read; if this non-digit is not a space, \TeX\ takes the token you did have
and backs it up, ready to be read again.  \ (On the other hand, the author
often omits the space when a constant is immediately followed by some
other character, because extra spaces do look funny in the file;
aesthetics are more important than~efficiency.)

\ddangerexercise Continuing the IRS example, assume that |\fullypaid|
and |\underpaid| are defined as follows:
\begintt
\def\fullypaid{Your taxes are fully paid---thank you.}
\def\underpaid{{\count0=-\balance
  \ifnum\count0<100
    You owe \dollaramount, but you need not pay it, because
    our policy is to disregard amounts less than \$1.00.
  \else Please remit \dollaramount\ within ten days,
    or additional interest charges will be due.\fi}}
\endtt
Write a macro |\overpaid| to go with these, assuming that |\dollaramount|
is a macro that generates the contents of\/ |\count0| in dollars and cents.
Your macro should say that a check will be mailed under separate cover,
unless the amount is less than \$1.00, in which case the person must
specifically request a check.
\answer |\def\overpaid{{\count0=\balance|\parbreak
        |  You have overpaid your tax by \dollaramount.|\parbreak
        |  \ifnum\count0<100 It is our policy to refund|\parbreak
        |    such a small amount only if you ask for it.|\parbreak
        |  \else A check for this amount is being mailed|\parbreak
        |    under separate cover.\fi}}|

\ddangerexercise Write a |\dollaramount| macro, to complete the Internal
Revenue |\statement|.
\answer The tricky part is to get the zero in an amount like `|$2.01|'.
\begintt
\def\dollaramount{\count2=\count0 \divide\count2 by100
  \$\number\count2.%
  \multiply\count2 by-100 \advance\count2 by\count0
  \ifnum \count2<10 0\fi
  \number\count2 }
\endtt

\ddanger Now let's make a complete survey of \TeX's conditional commands.
Some of them involve features that have not yet been introduced in this
manual.\enddanger

\nobreak\medskip
\item\bull^|\ifnum|\<number$_1$>\<relation>\<number$_2$>\quad
  (compare two integers)
\nobreak\smallskip\noindent
The ^\<relation> must be either `|<|$_{12}$' or `|=|$_{12}$' or `|>|$_{12}$'.
The two integer numbers are compared to each other in the usual way, and
the result is true or false accordingly.

\medbreak
\item\bull^|\ifdim|\<dimen$_1$>\<relation>\<dimen$_2$>\quad
  (compare two dimensions)
\nobreak\smallskip\noindent
This is like |\ifnum|, but it compares two \<dimen> values. For example,
to test whether the value of\/ |\hsize| exceeds $100\pt$, you can say
`|\ifdim\hsize>100pt|'.

\medbreak
\item\bull^|\ifodd|\<number>\quad(test for odd integer)
\nobreak\smallskip\noindent
The condition is true if the \<number> is odd, false if it is even.

\medbreak
\item\bull^|\ifvmode|\quad(test for vertical mode)
\nobreak\smallskip\noindent
True if \TeX\ is in vertical mode or internal vertical mode (see Chapter~13).

\medbreak
\item\bull^|\ifhmode|\quad(test for horizontal mode)
\nobreak\smallskip\noindent
True if \TeX\ is in horizontal mode or restricted horizontal mode
(see Chapter~13).

\medbreak
\item\bull^|\ifmmode|\quad(test for math mode)
\nobreak\smallskip\noindent
True if \TeX\ is in math mode or display math mode (see Chapter~13).

\medbreak
\item\bull^|\ifinner|\quad(test for an internal mode)
\nobreak\smallskip\noindent
True if \TeX\ is in internal vertical mode, or restricted
horizontal mode, or (non-display) math mode (see Chapter~13).

\medbreak
\item\bull^|\if|\<token$_1$>\<token$_2$>\quad(test if character codes agree)
\nobreak\smallskip\noindent
\TeX\ will expand macros following |\if| until two unexpandable tokens are
found. If either token is a control sequence, \TeX\ considers it to have
character code~256 and category code~16, unless the current equivalent of
that control sequence has been |\let| equal to a non-active character
token.  In this way, each token specifies a (character~code,
\hbox{category}~code) pair.  The condition is true if the character codes
are equal, independent of the category codes. For example, after
|\def\a{*}| and |\let\b=*| and |\def\c{/}|, the tests `|\if*\a|' and
`|\if\a\b|' will be true, but `|\if\a\c|' will be false. Also
`|\if\a\par|' will be false, but `|\if\par\let|' will be true. % Beresford=true

\medbreak
\item\bull^|\ifcat|\<token$_1$>\<token$_2$>\quad(test if category codes agree)
\nobreak\smallskip\noindent
This is just like |\if|, but it tests the ^{category codes}, not the character
codes. ^{Active characters} have category~13, but you have to say
`^|\noexpand|\<active character>' in order to suppress expansion when you
are looking at such characters with |\if| or |\ifcat|. For example, after
\begintt
\catcode`[=13 \catcode`]=13 \def[{*}
\endtt
the tests `|\ifcat\noexpand[\noexpand]|' and `|\ifcat[*|' will be true,
but the test `|\ifcat\noexpand[*|' will be false.

\medbreak
\item\bull^|\ifx|\<token$_1$>\<token$_2$>\quad(test if tokens agree)
\nobreak\smallskip\noindent
In this case, \TeX\ does {\sl not\/} expand control sequences when it
looks at the two tokens. The condition is true if (a)~the two tokens are
not macros, and they both represent the same (character code, category
code) pair or the same \TeX\ primitive or the same ^|\font| or ^|\chardef|
or ^|\countdef|, etc.; or if (b)~the two tokens are macros, and they both
have the same status with respect to ^|\long| and ^|\outer|, and they both
have the same parameters and ``top level'' expansion. For example, after
`|\def\a{\c}| |\def\b{\d}| |\def\c{\e}| |\def\d{\e}| |\def\e{A}|', an
|\ifx| test will find |\c| and |\d| equal, but not |\a| and~|\b|, nor |\d|
and~|\e|, nor any other combinations of\/ |\a|, |\b|, |\c|, |\d|, |\e|.

\medbreak
\item\bull^|\ifvoid|\<number>, ^|\ifhbox|\<number>, ^|\ifvbox|\<number>\quad
(test a box register)
\nobreak\smallskip\noindent
The \<number> should be between 0 and 255. The condition is true if that
|\box| is void or contains an hbox or a vbox, respectively (see Chapter~15).

\medbreak
\item\bull^|\ifeof|\<number>\quad(test for end of file)
\nobreak\smallskip\noindent
The \<number> should be between 0 and 15. The condition is true unless the
corresponding input stream is open and not fully read. \ (See the command
^|\openin| below.)

\medbreak
\item\bull^|\iftrue|, ^|\iffalse|\quad(always true or always false)
\nobreak\smallskip\noindent
These conditions have a predetermined outcome. But they turn out to be
useful in spite of this, as explained below.

\medbreak
Finally, there's one more conditional construction, which is somewhat different
from the rest because it is capable of making a many-way branch:
\begindisplay
\llap{\bull\enspace}^|\ifcase|\<number>\<text for case 0>|\or|\<text
  for case 1>|\or|$\;\cdots$\cr
|         \or|\<text for case $n$>|\else|\<text for all other cases>|\fi|\cr
\enddisplay
Here there are $n+1$ cases separated by $n$ ^|\or|'s, where $n$ can be any
nonnegative number. The \<number> selects the text that will be used. Once
again the ^|\else| part is optional, if you don't want to specify any text
for cases when the \<number> is negative or greater than~$n$.

\ddangerexercise Design a |\category| macro that prints a character's current
category code symbolically, given a one-character control sequence for
that character. For example, if the category codes of plain \TeX\ are in
force, `|\category\\|' should expand to `|escape|', and `|\category\a|'
should expand to `|letter|'.
\answer |\def\category#1{\ifcase\catcode`#1|\parbreak
        |  escape\or begingroup\or endgroup\or math\or|\parbreak
        |  align\or endline\or parameter\or superscript\or|\parbreak
        |  subscript\or ignored\or space\or letter\or|\parbreak
        |  otherchar\or active\or comment\or invalid\fi}|\par

\ddangerexercise Test yourself on the following questions to see if you
understand certain borderline situations: After the definitions
`|\def\a{}| |\def\b{**}| |\def\c{True}|',
which of the following are true?
(a)~`|\if\a\b|';
(b)~`|\ifcat\a\b|';
(c)~`|\ifx\a\b|';
(d)~`|\if\c|';
(e)~`|\ifcat\c|';
(f)~`|\ifx\ifx\ifx|'.
(g)~`|\if\ifx\a\b\c\else\if\a\b\c\fi\fi|'.
\answer (a,b)~True. (c,d)~False. (e,f)~True. In case~(e), the \<true text>
starts with `|ue|'.  (g)~The |\ifx| is false and the inner |\if| is true;
so the outer |\if| becomes `|\if True...|', which is false. \
(Interestingly, \TeX\ knows that the outer |\if| is false even before it
has looked at the |\fi|'s that close the |\ifx| and the inner |\if|.)

\ddanger Notice that all of the control sequences for conditionals
begin with |\if...|, and they all have a matching~|\fi|. This convention---that
|\if...|\ pairs up with |\fi|---makes it easier to see the nesting
of conditionals within your program. The ^{nesting} of\/ |\if...\fi| is
independent of the nesting of |{...}|; thus, you can begin or end a ^{group}
in the middle of a conditional, and you can begin or end a conditional
in the middle of a group. Extensive experience with macros has shown that
such independence is important in applications; but it can also lead to
confusion if you aren't careful.

\ddanger It's sometimes desirable to pass information from one macro to
another, and there are several ways to do this: ^^{communication between
macros} by passing it as an argument, by putting it into a register, or by
defining a control sequence that contains the information. For example,
the macros |\hphantom|, |\vphantom|, and ^|\phantom| in Appendix~B are
quite similar, so the author ^^{Knuth} wanted to do most of the work in
another macro |\phant| that would be common to all three. Somehow |\phant|
was to be told what kind of phantom was desired. The first approach was to
define control sequences |\hph| and |\vph| something like this:
\begintt
\def\hphantom{\ph YN} \def\vphantom{\ph NY} \def\phantom{\ph YY}
\def\ph#1#2{\def\hph{#1}\def\vph{#2}\phant}
\endtt
after which |\phant| could test `|\if Y\hph|' and `|\if Y\vph|'. This
worked, but there were various ways to make it more efficient; for example,
`|\def\hph{#1}|' could be replaced by `|\let\hph=#1|', avoiding macro expansion.
An even better idea then suggested itself:
\begintt
\def\yes{\if00} \def\no{\if01}
\def\hphantom{\ph\yes\no}...\def\phantom{\ph\yes\yes}
\def\ph#1#2{\let\ifhph=#1\let\ifvph=#2\phant}
\endtt
after which |\phant| could test `|\ifhph|' and `|\ifvph|'. \ (This
construction was tried before |\iftrue| and |\iffalse| were part of the
\TeX\ language.) \ The idea worked fine, so the author started to use
|\yes| and |\no| in a variety of other situations. But then one day a
complex conditional failed, because it contained an |\ifhph|-like test
inside another conditional:
\begintt
\if...  \ifhph...\fi  ... \else  ...  \fi
\endtt
Do you see the problem that developed? When the \<true text> of the
outermost conditional was executed, everything worked fine, because
|\ifhph| was either |\yes| or |\no| and it expanded into either |\if00| or
|\if01|.  But when the \<true text> was skipped, the |\ifhph| was not
expanded, so the first |\fi| was mistakenly paired with the first |\if|;
everything soon went haywire. That's when ^|\iftrue| and ^|\iffalse|
were put into the language, in place of\/ |\yes| and |\no|; now
|\ifhph| is either |\iftrue| or |\iffalse|, so \TeX\ will match it
properly with a closing~|\fi|, whether or not it is being skipped over.

\ddanger To facilitate |\if...|~constructions, plain \TeX\ has a
^|\newif| macro, such that after you say `|\newif\ifabc|' three
control sequences will be defined: |\ifabc| (for testing the switch),
|\abctrue| (for making the switch true), and |\abcfalse| (for making
it false). The |\phantom| problem is now solved in Appendix~B by writing
\begintt
\newif\ifhph \newif\ifvph
\def\hphantom{\hphtrue\vphfalse\phant}
\endtt
and with similar definitions of\/ |\vphantom| and |\phantom|. There is no
longer any need for a |\ph| macro; again |\phant| tests |\ifhph| and
|\ifvph|. Appendix~E contains other examples of conditionals created
by~|\newif|. New conditionals are initially false.

\ddanger Caution: Don't say anything like `|\let\ifabc=\iftrue|' in
conditional text. If \TeX\ skips over this command, it will think that
both |\ifabc| and |\iftrue| require a matching |\fi|, since the |\let| is
not being executed! Keep such commands buried inside macros, so that \TeX\
will see the `|\if...|'\ only when it is not skipping over the text that it
is reading.

\ddanger \TeX\ has 256 ``^{token list registers}'' called ^|\toks||0|
through |\toks255|, so that token lists can easily be shuffled around
without passing them through \TeX's reading apparatus.
There's also a ^|\toksdef| instruction so that, e.g.,
\begintt
\toksdef\catch=22
\endtt
makes |\catch| equivalent to |\toks22|. Plain \TeX\ provides a ^|\newtoks|
macro that allocates a new token list register; it is analogous to
|\newcount|. Token list registers behave like the ^{token list parameters}
|\everypar|, |\everyhbox|, |\output|, |\errhelp|, etc. To assign
a new value to a token list parameter or register, you say either
\begindisplay
\<token variable>|={|\<replacement text>|}|\cr
\llap{or }\<token variable>|=|\<token variable>\cr
\enddisplay
where ^\<token variable> means either a token list parameter or a
control sequence defined by |\toksdef| or |\newtoks|, or an
explicit register designation `|\toks|\<number>'.

\ddanger Everyone who makes extensive use of a powerful macro facility
encounters situations when the macros do surprising things. We have
already mentioned the possibility of setting |\tracingmacros=1|, in order
to see when \TeX\ expands macros and what arguments it finds. There's also
another helpful way to watch what \TeX\ is doing: If you set
^|\tracingcommands||=1|, \TeX\ will show every command that it executes,
as we saw in Chapter~13.  Furthermore, if you set
|\tracingcommands=2|, \TeX\ will show all conditional commands and
their outcomes, as well as the unconditional commands that are actually
performed or expanded. This diagnostic information goes into your log
file. You can also see it on your terminal, if you say
^|\tracingonline||=1|. \ (Incidentally, if you make |\tracingcommands|
greater than~2, you get the same information as when it equals~2.) \
Similarly, ^|\tracingmacros||=2| will trace |\output|, |\everypar|, etc.

\ddanger One way to understand the occasional strangeness of macro
operation is to use the tracing features just described, so that you can
watch what \TeX\ does in slow motion. Another way is to learn the rules for
how macros are expanded; we shall now discuss those rules.

\ddanger \TeX's mastication process converts your input to a long token
list, as explained in Chapter~8; and its digestive processes work
strictly on this token list. When \TeX\ encounters a control sequence in
the token list, it looks up the current meaning, and in certain cases
it will expand that token into a sequence of other tokens before
continuing to read. The expansion process applies to macros and to
certain other special primitives like |\number| and |\if| that we shall
consider momentarily. Sometimes, however, the ^{expansion} is not carried out;
for example, when \TeX\ is taking care of a |\def|, the \<control sequence>,
the \<parameter text>, and the \<replacement text> of that |\def| are
not subject to expansion. Similarly, the two tokens after |\ifx| are
never expanded. A complete list of occasions when tokens are not expanded
appears later in this chapter; you can use it for reference in an emergency.

\ddanger Now let's consider the control sequences that are expanded whenever
expansion has not been inhibited. Such control sequences
fall into several classes:\enddanger

\nobreak\medskip
\textindent\bull Macros. When a macro is expanded, \TeX\ first determines its
arguments (if any), as explained earlier in this chapter. Each argument
is a token list; the tokens are not expanded when they are being
accepted as arguments. Then \TeX\ replaces the macro and its arguments
by the replacement text.

\smallbreak
\textindent\bull Conditionals. When an |\if...| is expanded, \TeX\ reads
ahead as far as necessary to determine whether the condition is true or
false; and if false, it skips ahead (keeping track of\/ |\if...\fi| nesting)
until finding the |\else|, |\or|, or~|\fi| that ends the skipped text.
Similarly, when |\else|, |\or|, or~|\fi| is expanded, \TeX\ reads to the
end of any text that ought to be skipped. The ``expansion'' of a conditional
is empty. \ (Conditionals always reduce the number of tokens that are seen by
later stages of the digestive process, while macros usually increase the
number of tokens.)

\smallbreak
\textindent\bull ^|\number|\<number>. When \TeX\ expands |\number|, it reads
the \<number> that follows (expanding tokens as it goes); the final
expansion consists of the ^{decimal representation} of that number,
preceded by `|-|' if negative.

\smallbreak
\textindent\bull ^|\romannumeral|\<number>. This is like |\number|, but
the expansion consists of lowercase roman numerals. For example,
`|\romannumeral 1984|' produces `|mcmlxxxiv|'. The expansion is
empty if the number is zero or negative.

\smallbreak
\textindent\bull ^|\string|\<token>. \TeX\ first reads the \<token> without
expansion. If a control sequence token appears, its |\string| expansion
consists of the control sequence name (including ^|\escapechar| as an escape
character, if the control sequence isn't simply an active character).
Otherwise the \<token> is a character token, and its character code
is retained as the expanded result.

\smallbreak
\textindent\bull ^|\jobname|. The expansion is the name that \TeX\ has chosen
for this job. For example, if \TeX\ is putting its output on files
|paper.dvi| and |paper.log|, ^^|dvi| then |\jobname| expands to `|paper|'.

\smallbreak
\textindent\bull ^|\fontname|\<font>. The expansion is the external file name
corresponding to the given font; e.g., `|\fontname\tenrm|' might expand
to `|cmr10|' (five tokens). If the font is not being used at its design size,
the ``^{at size}'' also appears in the expansion.  A ^\<font> is either an
identifier defined by ^|\font|; or |\textfont|\<number>,
|\scriptfont|\<number>, or |\scriptscriptfont|\<number>; or |\font|, which
denotes the current font.

\smallbreak
\textindent\bull ^|\meaning|\<token>. \TeX\ expands this to the sequence of
characters that would be displayed on your terminal by the commands
`|\let\test=|\<token> |\show\test|'. For example, `|\meaning A|' usually
expands to `|the letter A|'; `|\meaning\A|' after `|\def\A#1B{\C}|'
expands to `|macro:#1B->\C |'.

\smallbreak
\textindent\bull ^|\csname||...|^|\endcsname|. When \TeX\ expands |\csname|
it reads to the matching |\endcsname|, expanding tokens as it goes;
only character tokens should remain after this expansion has taken place.
Then the ``expansion'' of the entire |\csname...\endcsname| text
will be a single control sequence token, defined to be like |\relax| if
its meaning is currently undefined.

\smallbreak
\textindent\bull ^|\expandafter|\<token>. \TeX\ first reads the token that
comes immediately after |\expandafter|, without expanding it;
let's call this token~$t$. Then \TeX\ reads the token that comes after~$t$
(and possibly more tokens, if that token has an argument), replacing it by
its expansion. Finally \TeX\ puts~$t$ back in front of that expansion.

\smallbreak
\textindent\bull ^|\noexpand|\<token>. The expansion is the token itself;
but that token is interpreted as if its meaning were `|\relax|' if it is a
control sequence that would ordinarily be expanded by \TeX's expansion rules.

\smallbreak
\textindent\bull ^|\topmark|, ^|\firstmark|, ^|\botmark|, ^|\splitfirstmark|,
and ^|\splitbotmark|. \kern-1.7pt % This saves an overfull box (March 27, 1983)
The expansion is the token list in the corresponding ``^{mark}'' register
(see Chapter~23).

\smallbreak
\textindent\bull ^|\input|\<filename>. The expansion is null; but \TeX\
prepares to read from the specified file before looking at any more
tokens from its current source.

\smallbreak
\textindent\bull ^|\endinput|. The expansion is null. The next time \TeX\ gets
to the end of an |\input| line, it will stop reading from the file containing
that line.

\smallbreak
\textindent\bull ^|\the|\<internal quantity>. The expansion is a list of
tokens representing the current value of one of \TeX's variables, as
explained below. For example, `|\the\skip5|' might expand into
`|5.0pt plus 2.0fil|' (17~tokens).

\ddanger The powerful |\the| operation has many subcases, so we shall discuss
them one at a time. A variety of internal numeric quantities
can be brought up front:\enddanger

\nobreak\medskip
\textindent\bull |\the|\<parameter>, where \<parameter> is the name of
one of \TeX's ^{integer parameters} (e.g., |\the\widowpenalty|),
^{dimension parameters} (e.g., |\the\parindent|), ^{glue parameters}
(e.g., |\the\leftskip|), or ^{muglue parameters} (e.g., |\the\thinmuskip|).

\smallbreak
\textindent\bull |\the|\<register>, where \<register> is the name of
one of \TeX's integer ^{registers} (e.g., |\the\count|\stretch|0|),
dimension registers (e.g., |\the\dimen169|), glue registers (e.g.,
|\the\skip255|), or muglue registers (e.g., |\the\muskip\count2|).

\smallbreak
\textindent\bull |\the|\<codename>\<8-bit number>, where \<codename>
stands for either ^|\catcode|, ^|\mathcode|, ^|\lccode|, ^|\uccode|,
^|\sfcode|, or ^|\delcode|. For example, |\the\mathcode`/| produces the
current (integer) math code value for a slash.

\smallbreak
\textindent\bull |\the|\<special register>, where \<special register> is
one of the integer quantities ^|\prevgraf|, ^|\deadcycles|, ^|\insertpenalties|,
^|\inputlineno|, ^|\badness|,
or ^|\parshape| (denoting only the number of lines of\/ |\parshape|); or
one of the dimensions ^|\pagetotal|, ^|\pagegoal|, ^|\pagestretch|,
^|\pagefilstretch|, ^|\pagefillstretch|, ^|\pagefilllstretch|, ^|\pageshrink|,
^|\pagedepth|. In horizontal modes you can also refer to a special integer,
|\the\spacefactor|; in vertical modes there's a special dimension,
|\the\prevdepth|.

\smallbreak
\textindent\bull |\the|^|\fontdimen|\<parameter number>\<font>.
This produces a dimension; for example, parameter~6 of a font is its
``^{em}'' value, so `|\the\fontdimen6\tenrm|' yields `|10.0pt|' (six tokens).

\smallbreak
\textindent\bull |\the|^|\hyphenchar|\<font>, |\the|^|\skewchar|\<font>.
These produce the corresponding integer values defined for the specified font.

\smallbreak
\textindent\bull |\the|^|\lastpenalty|, |\the|^|\lastkern|, |\the|^|\lastskip|.
These yield the amount of penalty, kerning, glue, or
muglue in the final item on the current list, provided that the item is a
penalty, kern, or glue, respectively; otherwise they yield `|0|' or `|0.0pt|'.

\smallbreak
\textindent\bull |\the|\<defined character>, where \<defined character> is
a control sequence that has been given an integer value with ^|\chardef| or
^|\mathchardef|; the result is that integer value, in decimal notation.

\ddanger In all of the cases listed so far, |\the| produces a result that
is a sequence of ^{ASCII} character tokens. Category code~12 (``other'') is
assigned to each token, except that character code~32 gets category~10
(``space''). The same rule is used to assign ^{category codes} to the
tokens produced by ^|\number|, ^|\romannumeral|, ^|\string|, ^|\meaning|,
^|\jobname|, and ^|\fontname|.

\ddanger There also are cases in which |\the| produces non-character tokens,
either a font identifier like |\tenrm|, or an arbitrary token list:

\textindent\bull |\the|\<font> produces a font identifier that selects the
specified font. For example, `|\the\font|' is a control sequence corresponding
to the current font.

\goodbreak\smallskip
\textindent\bull |\the|\<token variable> produces a copy of the token list that
is the current value of the variable. For example, you can expand
`|\the\everypar|' and `|\the\toks5|'.

\ddanger \TeX's primitive command `^|\showthe|' will display on your
terminal exactly what `|\the|' would produce in an expanded definition;
the expansion is preceded by `|> |' and followed by a period. For example,
`|\showthe\parindent|' will display
\begintt
> 20.0pt.
\endtt
if the plain \TeX\ paragraph indentation is being used.

\ddanger Here now is the promised list of all cases when expandable tokens
are not expanded. Some of the situations involve primitives that
haven't been discussed yet, but we'll get to them
eventually.  Expansion is suppressed at the following times:\enddanger

\nobreak\medskip
\item\bull When tokens are being deleted during ^{error recovery} (see
Chapter~6).

\smallskip
\item\bull When tokens are being skipped because conditional text is being
ignored.

\smallskip
\item\bull When \TeX\ is reading the arguments of a macro.

\smallskip
\item\bull When \TeX\ is reading a control sequence to be defined by
^|\let|, ^|\futurelet|, ^|\def|, ^|\gdef|, ^|\edef|, ^|\xdef|, ^|\chardef|,
^|\mathchardef|, ^|\countdef|, ^|\dimendef|, ^|\skipdef|, ^|\muskipdef|,
^|\toksdef|, ^|\read|, and ^|\font|.

\smallskip
\item\bull When \TeX\ is reading argument tokens for ^|\expandafter|,
^|\noexpand|, ^|\string|, ^|\meaning|, ^|\let|, ^|\futurelet|, ^|\ifx|,
^|\show|, ^|\afterassignment|, ^|\aftergroup|.

\smallskip
\item\bull When \TeX\ is absorbing the parameter text of a |\def|, |\gdef|,
|\edef|, or |\xdef|.

\smallskip
\item\bull When \TeX\ is absorbing the replacement text of a |\def| or
|\gdef| or ^|\read|; or the text of a ^{token variable} like ^|\everypar|
or ^|\toks||0|; or the token list for ^|\uppercase| or ^|\lowercase| or
^|\write|.  \ (The token list for |\write| will be expanded later, when it
is actually output to a file.)

\smallskip
\item\bull When \TeX\ is reading the preamble of an alignment, except after
a token for the primitive command |\span| or when reading the \<glue>
after ^|\tabskip|.

\smallskip
\item\bull Just after a token such as |$|$_3$ that begins math mode, to see if
another token of category 3 follows.

\smallskip
\item\bull Just after a |`|$_{12}$ token that begins an ^{alphabetic constant}.

\ddanger Sometimes you will find yourself wanting to define new macros
whose replacement text has been expanded, based on current conditions,
instead of simply copying the replacement text verbatim. \TeX\ provides
the ^|\edef| (expanded definition) command for this purpose, and also
^|\xdef| (which is equivalent to |\global\edef|). The general format
is the same as for |\def| and |\gdef|, but \TeX\ blindly expands the tokens
of the replacement text according to the expansion rules above. For
example, consider
\begintt
\def\double#1{#1#1}
\edef\a{\double{xy}} \edef\a{\double\a}
\endtt
Here the first |\edef| is equivalent to `|\def\a{xyxy}|' and the second is
equivalent to `|\def\a{xyxyxyxy}|'. All of the other kinds of expansion
will take place too, including conditionals; for example,
\begintt
\edef\b#1#2{\ifmmode#1\else#2\fi}
\endtt
gives a result equivalent to `|\def\b#1#2{#1}|' if \TeX\ is in math
mode at the time of the |\edef|, otherwise the result is equivalent to
`|\def\b#1#2{#2}|'.

\ddanger Expanded definitions that are made with |\edef| or |\xdef| continue
to expand tokens until only unexpandable tokens remain, except that
token lists produced by `^|\the|' are not expanded further. Furthermore
a token following `^|\noexpand|' will not be expanded, since its ability
to expand has been nullified. These two operations can be used to control
^^{expansion, avoiding} what gets expanded and what doesn't.

\ddanger Suppose, for example, that you
want to define |\a| to be equal to~|\b| (expanded) followed by~|\c|
(not expanded) followed by |\d| (expanded), assuming that |\b| and
|\d| are simple macros without parameters. There are two easy ways to do it:
\begintt
\edef\a{\b\noexpand\c\d}
\toks0={\c} \edef\a{\b\the\toks0 \d}
\endtt
And it's even possible to achieve the same effect without using either
|\noexpand| or |\the|; a reader who wants to learn more about \TeX's
expansion mechanism is encouraged to try the next three exercises.

\ddangerexercise Figure out a way to define |\a| as in the previous paragraph,
without using \TeX's primitives `|\noexpand|' and `|\the|'.
\answer One idea is to say
\begintt
\let\save=\c \let\c=0 \edef\a{\b\c\d} \let\c=\save
\endtt
because control sequences equivalent to characters are not expandable.
However, this doesn't expand occurrences of~|\c|
that might be present in the expansions of\/ |\b| and~|\d|. Another way,
which is free of this defect, is
\begintt
\edef\next#1#2{\def#1{\b#2\d}} \next\a\c
\endtt
(and it's worth a close look!).

\ddangerexercise Continuing the example of expansion avoidance, suppose that
you want to expand |\b| completely until only unexpandable tokens are left,
but you don't want to expand |\c| at all, and you want to expand |\d|
only one level. For example, after |\def\b{\c\c}| and |\def\c{*}| and
|\def\d{\b\c}| the goal would be to get the effect of\/ |\def\a{**\c\b\c}|.
How can such a partial expansion be achieved, using ^|\the|?
\answer |\toks0={\c} \toks2=\expandafter{\d}|\parbreak
        |\edef\a{\b\the\toks0 \the\toks2 }|
\smallskip\noindent
(Notice that ^|\expandafter| expands the token after the left brace here.)

\ddangerexercise Solve the previous exercise without |\the| or |\noexpand|.
\ (This is difficult.)
\answer The following shouldn't be taken too seriously, but it does work:
^^|\span|
\begintt
{\setbox0=\vbox{\halign{#{\c\span\d}\cr
      \let\next=0\edef\next#1{\gdef\next{\b#1}}\next\cr}}}
\let\a=\next
\endtt

\ddanger \TeX's primitive commands ^|\mark||{...}|, ^|\message||{...}|,
^|\errmessage||{...}|, ^|\special||{...}|, and ^|\write|\<number>|{...}|
all expand the token lists in braces almost exactly as |\edef| and |\xdef| do.
However, a macro parameter
character like~|#| should not be duplicated in such commands; you need
to say |##| within an |\edef|, but only |#| within a |\mark|. The
|\write| command is somewhat special, because its token list is first
read without expansion; expansion occurs later, when the tokens
are actually being written to a file.

\ddangerexercise Compare the following two definitions:
\begintt
\def\a{\iftrue{\else}\fi}
\edef\b{\iftrue{\else}\fi}
\endtt
Which of them yields an ^{unmatched left brace}? (This is tricky.)
\answer Neither one, although |\a| will behave like an unmatched left
brace when it is expanded. The definition of\/ |\b| is {\sl not complete},
because it expands to `|\def\b{{}|'; \TeX\ will continue to read ahead,
looking for another right brace, possibly discovering a runaway
definition! It's impossible to define a macro that has unmatched braces.
But you {\sl can\/} say |\let\a={|; Appendix~D discusses several
other ^{brace tricks}.

\def\rhead{Chapter \chapno: Definitions (aka Macros)% my little joke
  \gdef\rhead{Chapter \chapno: Definitions (also called Macros)}}

\ddanger \TeX\ has the ability to read individual lines of text from up
to~16 files at once, in addition to the files that are being |\input|.
To initiate reading such an auxiliary file, you should say
\begindisplay
^|\openin|\<number>|=|\<filename>
\enddisplay
where the \<number> is between 0 and 15. \ (Plain \TeX\ allocates
input stream numbers 0~through~15 with the ^|\newread| command, which
is analogous to |\newbox|.) \ In most installations
of \TeX, the extension `^|.tex|' will be appended to the file name,
as with ^|\input|, if no extension is given explicitly. If the file
cannot be found, \TeX\ will give no error message; it will simply consider
that the input stream is not open, and you can test this condition
with ^|\ifeof|. When you're done with a file, you can say
\begindisplay
^|\closein|\<number>
\enddisplay
and the file associated with that input stream number will be closed,
i.e., returned to its initial condition, if such a file was open.
To get input from an open file, you say
\begindisplay
^|\read|\<number>^|to|\<control sequence>
\enddisplay
and the control sequence is defined to be a parameterless macro whose
replacement text is the contents of the next line read from the
designated file. This line is converted to a token list, using the
procedure of Chapter~8, based on the current category codes.
Additional lines are read, if necessary, until an equal number of
left and right braces has been found. An empty line is implicitly appended
to the end of a file that is being |\read|. ^^{empty line at end of file}
If the \<number> is not between 0 and~15, or if no such file is open, or if
the file has ended, input will be from the terminal; \TeX\ will prompt
the user unless the \<number> is negative. The macro definition will be
local unless you say |\global\read|.

\ddanger For example, it's easy to have ^{dialogs with the user}, by
using |\read| together with the ^|\message| command (which
writes an expanded token list on the terminal and in the log file):
\begintt
\message{Please type your name:}
\read16 to\myname
\message{Hello, \myname!}
\endtt
The |\read| command in this case will print `|\myname=|' and it will wait
for a response; the response will be echoed on the log file.
The `|\myname=|' would have been omitted if `|\read16|' had been
`|\read-1|'.

\ddangerexercise The |\myname| example just given doesn't work quite right,
because the \<return> at the end of the line gets translated into a
space. Figure out how to fix that glitch.
\answer One way is to redefine |\catcode`\^^M=9| (ignored) just before
the |\read|, so that the \<return> will be ignored. Another solution is
to redefine ^|\endlinechar||=-1|, so that no character is put at the
end of the line. Or you could try to be tricky by stripping off the
space with macro expansion as follows:
\begintt
\def\stripspace#1 \next{#1}
\edef\myname{\expandafter\stripspace\myname\next}
\endtt
The latter solution doesn't work if the user types `|%|' at the end of
his~or her name, or if the name contains control sequences.

\ddangerexercise Continuing the previous example, define a macro
|\MYNAME| that contains the letters of\/ |\myname| all in ^{uppercase
letters}. For example, if\/ |\myname| expands to |Arthur|, |\MYNAME|
should expand to |ARTHUR|. Assume that |\myname| contains only
letters and spaces in its expansion.
\answer Here are two solutions:
\begintt
\def\next#1\endname{\uppercase{\def\MYNAME{#1}}}
\expandafter\next\myname\endname
|smallskip\edef\next{\def\noexpand\MYNAME{\myname}}
\uppercase\expandafter{\next}
\endtt

\ddanger Appendices B, D, and E contain numerous examples of how to make
macros do useful things. Let's close this chapter by presenting a few
examples that show how \TeX\ can actually be used as a primitive ^{programming}
language, if you want to achieve special effects, and if you don't care very
much about computer costs.

\ddanger Plain \TeX\ contains a |\loop...\repeat| construction, which works
like this: You say `^|\loop|~$\alpha$~|\if...|~$\beta$~|\repeat|', where
$\alpha$ and~$\beta$ are any sequences of commands, and where |\if...|\
is any conditional test (without a matching |\fi|). \TeX\ will first
do~$\alpha$; then if the condition is true, \TeX\ will do~$\beta$ and
repeat the whole process again starting with~$\alpha$. If the
condition ever turns out to be false, the loop will stop. For example,
here is a program that carries out a little dialog in which \TeX\ waits for
the user to type `|Yes|' or `|No|': ^^{repeating commands, see :loop}
\begintt
\def\yes{Yes } \def\no{No } \newif\ifgarbage
\loop\message{Are you happy? }
  \read-1 to\answer
  \ifx\answer\yes\garbagefalse % the answer is Yes
    \else\ifx\answer\no\garbagefalse % the answer is No
      \else\garbagetrue\fi\fi % the answer is garbage
  \ifgarbage\message{(Please type Yes or No.)}
\repeat
\endtt

\ddangerexercise Use the |\loop...\repeat| mechanism to construct a
general |\punishment| macro that repeats any given paragraph any given number of
times. For example,
\begintt
\punishment{I must not talk in class.}{100}
\endtt
should produce the results desired in exercise 20.\punishexno.
\answer (Here's a solution that also numbers the lines, so that the number of
repetitions is easily verifiable.
The only tricky part about this answer is the use of\/ ^|\endgraf|,
which is a substitute for |\par| because |\loop| is not a ^|\long| macro.)
\begintt
\newcount\n
\def\punishment#1#2{\n=0
  \loop\ifnum\n<#2 \advance\n by1
    \item{\number\n.}#1\endgraf\repeat}
\endtt

\let\plaind=\d
\newif\ifprime \newif\ifunknown
\newcount\n \newcount\p \newcount\d \newcount\a
\def\primes#1{2,~3% assume that #1 is at least 3
  \n=#1 \advance\n by-2 % n more to go
  \p=5 % odd primes starting with p
  \loop\ifnum\n>0 \printifprime\advance\p by2 \repeat}
\def\printp{, % we will invoke \printp if p is prime
  \ifnum\n=1 and~\fi % this precedes the last value
  \number\p \advance\n by -1 }
\def\printifprime{\testprimality \ifprime\printp\fi}
\def\testprimality{{\d=3 \global\primetrue
  \loop\trialdivision \ifunknown\advance\d by2 \repeat}}
\def\trialdivision{\a=\p \divide\a by\d
  \ifnum\a>\d \unknowntrue\else\unknownfalse\fi
  \multiply\a by\d
  \ifnum\a=\p \global\primefalse\unknownfalse\fi}
\ddanger The first thirty prime numbers are \primes{30}. You may not
find this fact very startling; but you may be surprised to learn that
the previous sentence was typeset by saying
\begintt
The first thirty prime numbers are \primes{30}.
\endtt
\TeX\ did all of the calculation by expanding the |\primes| macro, so the author
is pretty sure that the list of ^{prime numbers} given above is quite free
of typographic errors. Here is the set of macros that did it:
\begindisplay
^|\newif||\ifprime \newif\ifunknown % boolean variables|\cr
^|\newcount||\n \newcount\p \newcount\d \newcount\a % integer variables|\cr
|\def\primes#1{2,~3% assume that #1 is at least 3|\cr
|  \n=#1 \advance\n by-2 % n more to go|\cr
|  \p=5 % odd primes starting with p|\cr
|  \loop\ifnum\n>0 \printifprime\advance\p by2 \repeat}|\cr
|\def\printp{, % we will invoke \printp if p is prime|\cr
|  \ifnum\n=1 and~\fi % `and' precedes the last value|\cr
|  \number\p \advance\n by -1 }|\cr
|\def\printifprime{\testprimality \ifprime\printp\fi}|\cr
|\def\testprimality{{\d=3 \global\primetrue|\cr
|  \loop\trialdivision \ifunknown\advance\d by2 \repeat}}|\cr
\noalign{\penalty-500}
|\def\trialdivision{\a=\p \divide\a by\d|\cr
|  \ifnum\a>\d \unknowntrue\else\unknownfalse\fi|\cr
|  \multiply\a by\d|\cr
|  \ifnum\a=\p \global\primefalse\unknownfalse\fi}|\cr
\enddisplay
^^|\multiply|^^|\divide|^^|\advance|^^|\newcount|
The computation is fairly straightforward, except that it involves
a loop inside a loop; therefore |\testprimality| introduces an extra
set of braces, to keep the inner loop control from interfering with
the outer loop. The braces make it necessary to say `^|\global|' when
|\ifprime| is being set true or false.  \TeX\ spent more time constructing
that sentence than it usually spends on an entire page; the
|\trialdivision| macro was expanded 132 times. % cpu time was 4 sec

\let\d=\plaind
\ddanger The |\loop| macro that does all these wonderful things is
actually quite simple. It puts the code that's supposed to be repeated
into a control sequence called |\body|, and then another control
sequence iterates until the condition is false:
\begintt
\def\loop#1\repeat{\def\body{#1}\iterate}
\def\iterate{\body\let\next=\iterate\else\let\next=\relax\fi\next}
\endtt
The expansion of\/ |\iterate| ends with the expansion of\/ |\next|; therefore
\TeX\ is able to remove |\iterate| from its memory before
invoking |\next|, and the memory does not fill up during a long loop.
Computer scientists call this ``^{tail recursion}.''

\ddanger The |\hex| macro below, which converts count register |\n| to
^{hexadecimal notation}, illustrates a {\sl recursive\/} control structure
in which many copies of\/ |\hex| can be active simultaneously.
^{Recursion} works better than simple |\loop| ^{iteration} in this
application because the hexadecimal digits are discovered from right to
left, while they must be output from left to right.
\ (The number in |\n| should be $\ge0$.)
\begintt
\def\hex{{\count0=\n \divide\n by16
  \ifnum\n>0 \hex\fi \count2=\n \multiply\count2 by-16
  \advance\count0 by\count2 \hexdigit}}
\def\hexdigit{\ifnum\count0<10 \number\count0
  \else\advance\count0 by-10 \advance\count0 by`A \char\count0 \fi}
\endtt

\ddanger Our final example is a macro that computes the number of nonblank
tokens in its argument; for example, `|\length{argument}|' expands to
`|8|'. This illustrates yet another aspect of macro technique.
\begintt
\def\length#1{{\count0=0 \getlength#1\end \number\count0}}
\def\getlength#1{\ifx#1\end \let\next=\relax
  \else\advance\count0 by1 \let\next=\getlength\fi \next}
\endtt

\endchapter

By this time [37 A.D.] the influence of ^{Macro} had become supreme.
\author ^{TACITUS}, {\sl Annals\/} (c.\thinspace120 A.D.) % book VI, ch 45

\bigskip

% Oh, you want a definition.
I hate definitions.
\author BENJAMIN ^{DISRAELI}, {\sl Vivian Grey\/} (1826) % Book II, Chapter 6

\eject
\beginchapter Chapter 21. Making Boxes

In Chapters 11 and 12 we discussed the principles of boxes and glue, and by now
we have seen many applications of those concepts. You can get by in most cases
with the boxes that \TeX\ manufactures automatically with its paragraph
builder, its page builder, and its math formula processor; but if you want to
do nonstandard things, you have the option of making boxes by yourself.
For example, Chapter~14 points out that you can keep something from being
hyphenated or split between lines if you enclose it in an ^|\hbox|;
Chapter~19 points out that |\hbox| allows you to get ordinary text into
a displayed equation. ^^|\vbox|

\danger The purpose of the present chapter is to nail down whatever details
about boxes haven't been covered yet. Fortunately, there isn't much more
to discuss; we have already mentioned most of the rules, so this chapter is
fairly short. In fact, the previous chapters have dealt with almost
everything except the rules about rules.

\danger To make a ^{rule box} (i.e., a solid ^{black rectangle}), you
type `^|\hrule|' in vertical mode or `^|\vrule|' in horizontal mode,
followed by any or all of the specifications `^|width|\<dimen>',
`^|height|\<dimen>', `^|depth|\<dimen>', in any order. For example, if
\begintt
\vrule height4pt width3pt depth2pt
\endtt
appears in the middle of a paragraph, \TeX\ will typeset the black box
`\thinspace\vrule height4pt width3pt depth2pt\thinspace'. If you
specify a dimension twice, the second specification overrules the first.
If you leave a dimension unspecified, you get the following by default:
$$\halign{\indent#\hfil&\qquad\hfil#\hfil&\qquad\hfil#\hfil\cr
&|\hrule|&|\vrule|\cr
width&\tt*&$0.4\pt$\cr
height&$0.4\pt$&\tt*\cr
depth&$0.0\pt$&\tt*\cr}$$
Here `{\tt*}' means that the actual dimension depends on the context;
the rule will extend to the boundary of the smallest box or alignment
that encloses it.

\hrule

\danger For example, the author typed `|\hrule|' just before typing this
paragraph, and you can see what happened: A horizontal rule, $0.4\pt$
thick, was extended across the page, because the vertical box that
encloses it turned out to be just that wide. \ (In fact, the
vertical box that encloses it is the page itself.) \ Another example
appears immediately after this paragraph, where you can see the result
of
\begintt
\hrule width5cm height1pt \vskip1pt \hrule width6cm
\endtt
\TeX\ does not put ^{interline glue} between rule boxes and their neighbors
in a vertical list, so these two rules are exactly $1\pt$ apart.

\hrule width5cm height1pt \vskip1pt \hrule width6cm

\dangerexercise B. L. ^{User} didn't want one of his horizontal rules to
touch the left margin, so he put it in a box and moved it right, like this:
\begintt
\moveright 1in \vbox{\hrule width3in}
\endtt
^^|\moveright|
But he found that this produced more space above and below the rule
than when he had simply said `|\hrule width 4in|' with no |\vbox|.
Why did \TeX\ insert more space, and what should he have done to
avoid it?
\answer The interline glue is added for vboxes, but not for rules; he
forgot to say ^|\nointerlineskip|, before and after the |\moveright|
construction.

\danger If you specify all three dimensions of a rule, there's no essential
difference between |\hrule| and |\vrule|, since both will produce exactly
the same black box. But you must call it an |\hrule| if you want to put
it in a vertical list, and you must call it a |\vrule| if you want to put
it in a horizontal list, regardless of whether it actually looks like
a horizontal rule or a vertical rule or neither. If you say |\vrule| in
vertical mode, \TeX\ starts a new paragraph; if you say |\hrule| in
horizontal mode, \TeX\ stops the current paragraph and returns to
vertical mode.

\danger The dimensions of a rule can be negative; for example, here's
a rule whose height is $3\pt$ and whose depth is $-2\pt$:
`\thinspace\vrule height3pt depth-2pt width1in\thinspace'.
However, a rule is invisible unless its height plus depth is positive
and its width is positive. A rule whose width is negative cannot be
seen, but it acts like a ^{backspace} when it appears in a horizontal list.

\dangerexercise Explain how the author probably obtained the rule
`\thinspace\vrule height3pt depth-2pt width1in\thinspace' in the
previous paragraph. [{\sl Hint:\/} It's one inch long.]
\answer |\vrule height3pt depth-2pt width1in|. Notice that it was necessary
to call it a |\vrule| since it appeared in horizontal mode.

\ddanger Now let's summarize all of the ways there are to specify boxes
explicitly to \TeX. \ (1)~A character by itself makes a character box, in
horizontal mode; this character is taken from the current font. \
(2)~The commands |\hrule| and |\vrule| make rule boxes, as just explained.
\ (3)~Otherwise you can make hboxes and vboxes, which fall under the
generic term ^\<box>. A \<box> has one of the following seven forms:
$$\halign{\indent#\hfil&\quad(see Chapter #)\hfil\cr
^|\hbox|\<box specification>|{|\<horizontal mode material>|}|&12\cr
^|\vbox|\<box specification>|{|\<vertical mode material>|}|&12\cr
^|\vtop|\<box specification>|{|\<vertical mode material>|}|&12\cr
^|\box|\<register number>&15\cr
^|\copy|\<register number>&15\cr
^|\vsplit|\<register number>|to|\<dimen>&15\cr
^|\lastbox|&21\cr}$$
Here a ^\<box specification> is either `^|to|\<dimen>' or
`^|spread|\<dimen>' or empty; this governs the setting of glue in
the horizontal or vertical lists inside the box, as explained in
Chapter~12. A \<register number> is between 0 and 255; after you say
|\box|, that register becomes void, but after |\copy| the register
is unchanged, as explained in Chapter~15. The |\vsplit| operation
is also explained in Chapter~15. In math modes an additional
type of box is available:
^|\vcenter|\<box specification>|{|\<vertical mode material>|}| (see Chapter~17).

\ddanger The bottom line of the table above refers to ^|\lastbox|,
a primitive operation that hasn't been mentioned before. If the last item
on the current horizontal list or vertical list is an hbox or vbox, it
is removed from the list and it becomes the |\lastbox|; otherwise
|\lastbox| is void.  This operation is allowed in ^{internal vertical
mode}, horizontal mode, and restricted horizontal mode, but you cannot use
it to take a box from the current page in vertical mode. In math modes,
|\lastbox| is always void. At the beginning of a paragraph,
`|{\setbox0=\lastbox}|' removes the ^{indentation box}.

\ddanger The operation ^|\unskip| is something like |\lastbox|, except that
it applies to glue instead of to boxes. If the last thing on the current list
is a glue item (or ^{leaders}, as explained below), it is removed. You
can't remove glue from the current page by using |\unskip| in vertical
mode, but you can say `|\vskip-|^|\lastskip|', which has almost
the same effect.

\ddanger Chapters 24 to 26 present summaries of all \TeX's operations
in all modes, and when those summaries mention a `\<box>' they mean one
of the seven possibilities just listed. For example, you can say
`|\setbox|\<register number>|=|\<box>' in any mode, and you can say
`|\moveright|\<dimen>\<box>' in vertical modes. But you can't say
`|\setbox|\<register number>|=C|' or `|\moveright|\<dimen>|\hrule|';
if you try either of these, \TeX\ will complain that a \<box> was
supposed to be present. Characters and rules are so special, they
aren't regarded as \<box>es.

\ddangerexercise Define a control sequence |\boxit| so that
`|\boxit{|\<box>|}|' yields the given box surrounded by 3~points of space and
by ruled lines on all four sides.
$$\def\boxit#1{\vbox{\hrule\hbox{\vrule\kern3pt
      \vbox{\kern3pt#1\kern3pt}\kern3pt\vrule}\hrule}}
\setbox4=\vbox{\hsize 23pc \noindent \strut For example, the sentence you
are now reading was typeset as part of the displayed formula
|$$\boxit{\boxit{\box4}}$$|, where box~4 was created by typing
`|\setbox4=\vbox{\hsize 23pc \noindent \strut For example,
the sentence you are now reading ... \strut}|'.\strut}
\boxit{\boxit{\box4}}$$
\answer |\def\boxit#1{\vbox{\hrule\hbox{\vrule\kern3pt|\parbreak
        |      \vbox{\kern3pt#1\kern3pt}\kern3pt\vrule}\hrule}}|\par
\smallskip\noindent
(The resulting box does not have the baseline of the original one;
you have to work a little bit harder to get that.)

\danger Let's look also at what can go inside a box. An hbox contains
a horizontal list; a vbox contains a vertical list. Both kinds of lists
are made up primarily of things like boxes, glue, kerns, and penalties,
as we have seen in Chapters 14 and~15. But you can also include some
special things that we haven't discussed yet, namely ``leaders''
and ``whatsits.'' Our goal in the rest of this chapter will be to
study how to make use of such exotic items.

\danger The dots you see before your eyes here
\xleaders\hbox to 6pt{$\hss\cdot\hss$}\hfil\
are called {\sl^{leaders}\/}\break
because they lead your eyes across the page; such things are often
used in indexes or tables of contents. The general idea is to repeat a
box as many times as necessary to fill up some given space. \TeX\ treats
leaders as a special case of glue; no, wait, it's the other way
around: \TeX\ treats glue as a special case of leaders. Ordinary glue
fills space with nothing, while leaders fill space with any
desired thing. In horizontal mode you can say
\begindisplay
^|\leaders|\<box or rule>|\hskip|\<glue>
\enddisplay
and the effect will be the same as if you had said just `|\hskip|\<glue>',
except that the space will be occupied by copies of the specified
\<box or rule>. The glue stretches or shrinks in the usual way. For example,
\begintt
\def\leaderfill{\leaders\hbox to 1em{\hss.\hss}\hfill}
\line{Alpha\leaderfill Omega}
\line{The Beginning\leaderfill The Ending}
\endtt
will produce the following two lines:
$$\vbox{
\def\leaderfill{\leaders\hbox to 1em{\hss.\hss}\hfill}
\line{Alpha\leaderfill Omega}
\line{The Beginning\leaderfill The Ending}
}$$
Here `|\hbox to 1em{\hss.\hss}|' specifies a box one em wide, with a
period in its center; the control sequence |\leaderfill| then causes this
box to be replicated when filling space in the |\line| box. \ (Plain \TeX's
^|\line| macro makes an hbox whose width is the |\hsize|.)

\danger Notice that the ^{dots} in the two example lines appear exactly
above each other. This is not a coincidence; it's a consequence of the
fact that the |\leaders| operation acts something like a window that lets
you see part of an infinite row of boxes. If the words `Alpha' and `Omega'
are replaced by longer words, the number of dots might be different
but the ones that you see will be in the same places as before. The infinitely
replicated boxes are lined up so that they touch each other, and so that,
if you could see them all, one of them would have the same reference point as
the smallest enclosing box. Thus, |\leaders| will put a box flush with the
left edge of an enclosing box, if you start the leaders there; but you
won't get a box flush right unless the width of the enclosing box is
exactly divisible by the width of the repeated box. If~the repeated box
has width~$w$, and if the space to be filled is at least~$2w$, then you
will always see at least one copy of the box; but if the space is less
than~$2w$ the box may not appear, because boxes in the infinite
row are typeset only when their entire width falls into the available space.

\ddanger When leaders are isolated from each other, you might not
want them to be aligned as just described, so \TeX\ also provides
for ^{nonaligned leaders}. In this case a box of width~$w$ will be
copied $q$~times when the space to be filled is at least~$qw$ and
less than $(q+1)w$; furthermore, the results will be centered in the
available space. There are two kinds of nonaligned leaders in \TeX,
namely ^|\cleaders| (centered leaders) and ^|\xleaders| (expanded
leaders). Centered leaders pack the boxes tightly next to each other
leaving equal amounts of blank space at the left and right; expanded
leaders distribute the extra space equally between the $q+1$ positions
adjacent to the $q$~boxes. For example, let's suppose that a $10\pt$-wide
box is being used in leaders that are supposed to fill $56\pt$ of space.
Five copies of the box will be used; |\cleaders| will produce $3\pt$ of
space, then the five boxes, then another $3\pt$ of space. But
|\xleaders| will produce $1\pt$ of space, then the first box, then
another $1\pt$ of space, then the second box, \dots, then the fifth box,
and $1\pt$ of space.

\ddangerexercise Suppose that a $10\pt$-wide box is to fill $38\pt$ of
space starting $91\pt$ from the left of its enclosing box. How many
copies of the box will be produced by |\leaders|, |\cleaders|, and
|\xleaders|? Where will the boxes be positioned, relative to the
left edge of the enclosing box, in each of the three cases?
\answer |\leaders|: two boxes starting at $100\pt$, $110\pt$.\par
|\cleaders|: three boxes starting at $95\pt$, $105\pt$, $115\pt$.\par
|\xleaders|: three boxes starting at $93\pt$, $105\pt$, $117\pt$.

\ddangerexercise Assuming that the `|.|'\ in the |\leaderfill| macro on the
previous page is only $0.2\em$ wide, there is $0.4\em$ of blank space
at both sides of the one-em box. Therefore the |\leaders| construction
will leave between $0.4\em$ and $1.4\em$ of blank space between the
periods and the text at either end. Redefine |\leaderfill| so that
the amount of blank space at either end will be between $0.1\em$ and
$1.1\em$, but the leaders on adjacent lines will still be aligned
with each other.
\answer |\def\leaderfill{\kern-0.3em\leaders\hbox to 1em{\hss.\hss}%|\parbreak
        |    \hskip0.6em plus1fill \kern-0.3em }|

\danger Instead of giving a \<box> in the leaders construction, you can
give a \<rule>, which means either |\hrule| or |\vrule|, followed
by optional ^|height|, ^|width|, and ^|depth| specifications as
usual. The rule will then be made as wide as the corresponding~\<glue>.
This is a case where |\hrule| makes sense in horizontal mode, because
it gives a ^{horizontal rule in text}. For example, if the
|\leaderfill| macro in our earlier illustration is changed to
\begintt
\def\leaderfill{ \leaders\hrule\hfill\ }
\endtt
then the results look like this:
$$\vbox{
\def\leaderfill{ \leaders\hrule\hfill\ }
\line{Alpha\leaderfill Omega}
\line{The Beginning\leaderfill The Ending}
}$$
When a rule is used instead of a box, it fills the space completely, so
there's no difference between |\leaders|, |\cleaders|, and |\xleaders|.

\ddangerexercise What does |\leaders\vrule\hfill| produce?
\answer Since no |height| or |depth| specification follows the |\vrule|,
the height and depth are `|*|'; i.e., the rule extends to the smallest
enclosing box. This usually makes a heavy black band, which is too
horrible to demonstrate here. However, it does work if you know that the
enclosing box is sufficiently small; and |\leaders\vrule\vfill| works fine in
vertical mode.

\ddanger Leaders work in vertical mode as well as in horizontal mode. In this
case vertical glue (e.g., |\vskip|\<glue> or |\vfill|) is used instead of
horizontal glue, and |\leaders| produces boxes that are aligned so that the
top of each repeated box has the same vertical position as the top
of the smallest enclosing box, plus a multiple of the height-plus-depth
of the repeated box. No interline glue is placed between boxes in
vertical leaders; the boxes are just stacked right on top of each other.

\ddanger If you specify horizontal leaders with a box whose width isn't
positive, or if you specify vertical leaders with a box whose
height-plus-depth isn't positive, \TeX\ silently ignores the leaders
and produces ordinary glue instead.

\ddangerexercise Explain how you can end a paragraph with a rule that
is at least $10\pt$ long and extends all the way to the right
margin, like this:
\null\nobreak\leaders\hrule\hskip10pt plus1filll\ \par
\answer For example, say this:
\begintt
\null\nobreak\leaders\hrule\hskip10pt plus1filll\ \par
\endtt
The `|\|\]' provides extra glue that is wiped out by the implied |\unskip|
at the end of every paragraph (see Chapter~14), and the `|\null\nobreak|'
makes sure that the leaders do not disappear at a line break; `^|filll|'
overtakes the ^|\parfillskip| glue.

\ddanger Horizontal leaders differ slightly from horizontal glue, because
they have height and depth when \TeX\ calculates the size of the enclosing
box (even though the number of replications might be zero).  Similarly,
vertical leaders have width.

\ddangerexercise Demonstrate how to produce the following `\TeX ture'
$$\hbox to 2.5in{\cleaders\vbox to .5in{\cleaders\hbox{\TeX}\vfil}\hfil}$$
by using vertical leaders inside of horizontal leaders. \ (The \TeX\ logo
has been put into a rectangular box, and copies of this box have been packed
together tightly.) ^^{TeX logo}
\answer |$$\hbox to 2.5in{\cleaders|\parbreak
        |    \vbox to .5in{\cleaders\hbox{\TeX}\vfil}\hfil}$$|

\ddangerexercise Use vertical leaders to solve exercise 20.\punishexno.
\answer We assume that a strut is $12\pt$ tall, and that 50 lines
fit on a page:
\begintt
\setbox0=\hbox{\strut I must not talk in class.}
\null\cleaders\copy0\vskip600pt\vfill\eject % 50 times on page 1;
\null\cleaders\box0\vskip600pt\bye % 50 more on page 2.
\endtt
The ^|\null| keeps glue (and leaders) from disappearing at the top of
the page.

\danger The ^|\overbrace| and ^|\underbrace| macros of plain \TeX\ are
constructed by combining characters with rules. Font ^|cmex10| contains
four symbols $\bracelu\;\braceld\;\bracerd\;\braceru$, each of which has
depth zero and height equal to the thickness of a rule that joins them
properly. Therefore it's easy to define ^|\upbracefill| and ^|\downbracefill|
macros so that you can obtain, e.g., ^^{braces}
\begindisplay
\vbox{\kern4pt
  \hbox to 100pt{\downbracefill}\hbox to 50pt{\upbracefill}
  }
\enddisplay
by saying `|\hbox to 100pt{\downbracefill}\hbox to 50pt{\upbracefill}|'
in vertical mode. Details of those macro definitions appear in Appendix~B.

\ddanger The definition of\/ ^|\overrightarrow| in Appendix B is more complex
than that of\/ |\overbrace|, because it involves a box instead of a rule. The
fonts of plain \TeX\ are designed so that symbols like $\leftarrow$ and
$\to$ can be extended with ^{minus signs}; similarly, $\Leftarrow$ and
^^{arrows}
$\Rightarrow$ can be extended with ^{equals signs}. However, you can't
simply put the characters next to each other, because that leaves
gaps (`${\leftarrow}{-}{-}$' and `${\Leftarrow}{=}{=}$'); it is necessary
to backspace a little between characters. An additional complication
arises because the extension line in a long arrow might need to be some
non-integer number of minus signs long. To solve this problem, the
^|\rightarrowfill| macro in Appendix~B uses |\cleaders| with a repeatable
box consisting of the middle 10 units of a minus sign, where one unit
is ${1\over18}\em$. The leaders are preceded and followed by $-$ and~$\to$;
there's enough backspacing to compensate for up to 5~units of extra space,
fore and aft, that |\cleaders| might leave blank. In this way a macro is
obtained such that
\begintt
\hbox to 100pt{\rightarrowfill}
\endtt
yields `\hbox to 100pt{\rightarrowfill}'.

\ddanger Now we know all about leaders. What about ^{whatsits}? Well,
whatsits have been provided as a general mechanism by which important
special printing applications can be handled as extensions to \TeX\null.
It's possible for system wizards to modify the \TeX\ program, without
changing too much of the code, so that new features can be accommodated
at high speed instead of encoding them in macros. The author hopes that
such extensions will not be made very often, because he doesn't want
incompatible pseudo-\TeX\ systems to proliferate; yet he realizes that
certain special books deserve a special treatment. Whatsits make it
possible to incorporate new things into boxes without bending the existing
conventions too much. But they make applications less portable from
one machine to another.

\ddanger Two kinds of whatsits are defined as part of all \TeX\
implementations.  They aren't really ^{extensions to \TeX}, but
they are coded as if they were, so that they provide a model of how
other extensions could be made. The first of these is connected with
output to text files, and it involves the \TeX\ primitive commands
^|\openout|, ^|\closeout|, ^|\write|, and ^|\immediate|. The second is
connected with special instructions that can be transmitted to
printing devices, via \TeX's ^|\special| command.

\ddanger The ability to write text files that can later be input by other
programs (including \TeX) makes it possible to take care of tables of
contents, indexes, and many other things. You can say
`|\openout|\<number>|=|\<filename>' and `|\closeout|\<number>' by analogy
with the |\openin| and |\closein| commands of Chapter~20; the
\<number> must be between 0 and~15. The file name is usually extended with
`^|.tex|' if it has no extension. There is a |\write| command that writes
one line to a file, analogous to the |\read| command that reads one line;
you say
\begindisplay
|\write|\<number>|{|\<token list>|}|
\enddisplay
and the material goes out to the file that corresponds to the given stream
number.  If the \<number> is negative or greater than~15, or if the
specified stream has no file open for output, the output goes to the
user's ^{log file}, and to the ^{terminal} unless the number is negative.
Plain \TeX\ has a ^|\newwrite| command that allocates output stream numbers
from 0 to~15. Output streams are completely independent of input streams.

\ddanger However, the output actually takes place in a delayed fashion;
the |\openout|, |\closeout|, and |\write| commands that you give are not
performed when \TeX\ sees them. Instead, \TeX\ puts these commands into
whatsit items, and places them into the current horizontal
or vertical or math list that is being built. No actual output will occur
until this whatsit is eventually shipped out to the |dvi| file, as part
of a larger box. The reason for this delay is that |\write| is often
used to make an index or table of contents, and the exact page on which
a particular item will appear is generally unknown when the |\write|
instruction occurs in mid-paragraph. \TeX\ is usually working ahead,
reading an entire paragraph before breaking it into lines, and accumulating
more than enough lines to fill a page before deciding what goes on
the page, as explained in Chapters 14 and~15. Therefore a deferred writing
mechanism is the only safe way to ensure the validity of page number
references.

\ddanger The \<token list> of a |\write| command is first stored in a
whatsit without performing any macro expansion; the macro expansion takes
place later, when \TeX\ is in the middle of a ^|\shipout| operation. For
example, suppose that some paragraph in your document contains the text
\begintt
... For \write\inx{example: \the\count0}example, suppose ...
\endtt
Then the horizontal list for the paragraph will have a whatsit just before
the word `|example|', and just after the interword space following~`|For|'.
This whatsit item contains the unexpanded token list
`|example: \the\count0|'. It sits dormant while the paragraph is being
broken into lines and put on the current page. Let's suppose that
this word `|example|' (or some hyphenated initial part of it, like
`|ex-|') is shipped out on page~256. Then \TeX\ will write the line
\begintt
example: 256
\endtt
on output stream |\inx|, because the `|\the\count0|' will be expanded
at that time. Of course, |\write| commands are usually generated by
macros; they are rarely typed explicitly in mid-paragraph.

\ddanger \TeX\ defers |\openout| and |\closeout| commands by putting them
into whatsits too; thus, the relative order of output commands will be
preserved, unless boxes are shipped out in some other order due to
insertions or such things.

\ddanger Sometimes you don't want \TeX\ to defer a |\write| or
|\openout| or |\closeout|. You could say, e.g., `|\shipout\hbox{\write...}|',
but that would put an unwanted empty page in your |dvi| file. So \TeX\ has
another feature that gets around this problem: If you type `^|\immediate|'
just before |\write| or |\openout| or |\closeout|, the operation will
be performed immediately, and no whatsit will be made. For example,
\begintt
\immediate\write16{Goodbye}
\endtt
prints `|Goodbye|' on your terminal. Without the |\immediate|, you wouldn't
see the `|Goodbye|' until the current list was output. \ (In fact, you
might never see it; or you may see it more than once, if the
current list goes into a box that was copied.) \ An `|\immediate\write16|'
differs from ^|\message| in that |\write| prints the text on a line by itself;
the results of several |\message| commands might appear on the same line,
separated by spaces.

\ddanger The \<token list> of a |\write| ought to be rather short, since
it makes one line of output. Some implementations of \TeX\ are unable to
write long lines; if you want to write a lot of stuff, just give
several |\write| commands. Alternatively, you can set \TeX's ^|\newlinechar|
parameter to the ASCII code number of some character that you want to
stand for ``begin a new line''; then \TeX\ will begin a new line whenever it
would ordinarily output that character to a file.
For example, one way to output two lines to the terminal
in a single |\write| command is to say
\begintt
\newlinechar=`\^^J
\immediate\write16{Two^^Jlines.}
\endtt
Each |\write| command produces output in the form that \TeX\ always uses
to display ^{token lists symbolically}: Characters represent
themselves (except that you get duplicated characters like |##| for macro
parameter characters^^{hash}); unexpandable control sequence tokens
produce their names, preceded by the ^|\escapechar| and followed by a
space (unless the name is an active character or a control sequence formed
from a single nonletter).

\ddanger \TeX\ ignores |\write|, |\openout|, and |\closeout| whatsits that
appear within boxes governed by ^{leaders}. If you are upset about this,
you shouldn't be.

\ddanger Since the \<token list> of a deferred |\write| is expanded at a
fairly random time (when |\shipout| occurs), you should be careful about
what control sequences it is allowed to contain. The techniques of Chapter~20
for controlling macro expansion often come in handy with respect to |\write|.

\ddangerexercise Suppose that you want to |\write| a token list that
involves a macro |\chapno|, containing the current chapter number, as well as
`|\the\count0|' which refers to the current page. You want |\chapno| to
be expanded immediately, because it might change before the token list
is written; but you want |\the\count0| to be expanded at the time of
|\shipout|. How can you manage this?
\answer If you say `|{\let|\stretch|\the=0\edef|\stretch|\next|\stretch
|{\write|\stretch|\cont|\stretch|{|\<token list>|}}\next|\stretch|}|',
the |\write| will be exercuted after |\edef| expands everything except |\the|.

\ddanger Now let's wrap up our study of boxes by considering one more feature.
The command `^|\special||{|\<token list>|}|' can be given in any
mode. Like |\write|, it puts its token list into a whatsit; and like
|\message|, it expands the token list immediately. This token list will be
output to the |dvi| file with the other typesetting commands that
\TeX\ produces. Therefore it is implicitly associated with a particular
position on the page, namely the reference point that would have been present
if a box of height, depth, and width zero had appeared in place of the
whatsit. The \<token list> in a |\special| command should consist of a
keyword followed if necessary by a space and appropriate arguments. For example,
\begintt
\special{halftone pic1}
\endtt
might mean that a ^{picture} on file |pic1| should be inserted ^^{halftones}
on the current page, with its reference point at the current position.
\TeX\ doesn't look at the token list to see if it makes any sense; the
list is simply copied to the output. However, you should be careful not to
make the list too long, or you might overflow \TeX's string memory.
The |\special| command enables you to make use of special equipment that
might be available to you, e.g., for printing books in glorious
\TeX ni^{color}.

\ddanger Software programs that convert |dvi| files to printed or
displayed output should be able to fail gracefully when they don't
recognize your special keywords. Thus, |\special| operations should
never do anything that changes the current position. Whenever you use
|\special|, you are taking a chance that your output file will not be
printable on all output devices, because all |\special| functions are
extensions to \TeX. However, the author anticipates that certain standards
for common graphic operations will emerge in the \TeX\ user community,
after careful experiments have been made by different groups of people;
then there will be a chance for some uniformity in the use of
|\special| extensions.

\ddanger \TeX\ will report the badness of glue setting in a box if
you ask for the numeric quantity
^|\badness| after making a box. For example, you might say
\begintt
\setbox0=\line{\trialtexta}
\ifnum\badness>250 \setbox0=\line{\trialtextb}\fi
\endtt
The badness is between 0 and 10000
unless the box is overfull, when |\badness=1000000|.


\endchapter

If age or weaknes doe prohibyte bloudletting,
you must use boxing.
\author PHILIP ^{BARROUGH}, {\sl The Methode of Phisicke\/} (1583) % page 6

\bigskip

The only thing that never looks right is a rule.
There is not in existence a page with a rule on it
that cannot be instantly and obviously improved
by taking the rule out.
% "Even dashes, cherished as they are by authors who cannot punctuate,
% spoil a page. They are generally merely ignorant substitutes for colons."
\author GEORGE BERNARD ^{SHAW}, in {\sl The Dolphin\/} (1940) % v4.1 p81

\eject
\beginchapter Chapter 22. Alignment

Printers charge extra when you ask them to typeset ^{tables}, and they do so
for good reason: Each table tends to have its own peculiarities, so it's
necessary to give some thought to each one, and to fiddle with alternative
approaches until finding something that looks good and communicates well.
However, you needn't be too frightened of doing tables with \TeX, since plain
\TeX\ has a ``tab'' feature that handles simple situations pretty much like
you would do them on a typewriter. Furthermore, \TeX\ has a powerful
alignment mechanism that makes it possible to cope with extremely complex
tabular arrangements. Simple cases of these ^{alignment} operations will
suffice for the vast majority of applications.

Let's consider ^{tabbing} first. If you say `^|\settabs| $n$ ^|\columns|',
plain \TeX\ makes it easy to produce lines that are divided into $n$~equal-size
^{columns}. Each line is specified by typing
\begindisplay
|\+|\<text$_1$>|&|\<text$_2$>|&|$\,\cdots\,$|\cr|
\enddisplay
where \<text$_1$> will start flush with the left margin,
\<text$_2$> will start at the left of the second column, and so on. Notice
that `^|\+|' starts the line. The final column is followed by `^|\cr|',
which old-timers will recognize as an abbreviation for the ``^{carriage
return}'' operation on typewriters that had carriages. For example,
consider the following specification:
\begintt
\settabs 4 \columns
\+&&Text that starts in the third column\cr
\+&Text that starts in the second column\cr
\+\it Text that starts in the first column, and&&&
  the fourth, and&beyond!\cr
\endtt
After `|\settabs|\stretch|4\columns|' each |\+| line is divided into quarters,
so the result~is
\medskip
\settabs 4 \columns
\+&&Text that starts in the third column\cr
\+&Text that starts in the second column\cr
\+\it Text that starts in the first column, and&&&
  the fourth, and&beyond!\cr
\def\tick{\kern-0.2pt % that's half the rule width
  \vbox to 0pt{\kern-36pt\leaders\hbox{\vrule height1pt\vbox to4pt{}}\vfil}}
\vskip-\baselineskip
\+\tick&\tick&\tick&\tick&\tick\cr
\medskip

This example merits careful study because it illustrates several things.
(1)~The `|&|' ^^{ampersand} is like the {\sc TAB} key on many typewriters;
it tells \TeX\ to advance to the next tab position, where there's a tab at
the right edge of each column. In this example, \TeX\ has set up four tabs,
indicated by the dashed lines; a dashed line is also shown at the left
margin, although there isn't really a tab there. (2)~But `|&|' isn't
exactly like a mechanical typewriter {\sc TAB}, because it first backs up
to the beginning of the current column before advancing to the next. In
this way you can always tell what column you're tabbing to, by counting
the number of |&|'s; that's handy, because variable-width type otherwise
makes it difficult to know whether you've passed a tab position or not.
Thus, on the last line of our example, three |&|'s were typed in order to
get to column~4, even though the text had already extended into column~2
and perhaps into column~3. (3)~You can say `|\cr|' before you have
specified a complete set of columns, if the remaining columns are blank.
(4)~The |&|'s are different from tabs in another way, too: \TeX\ ignores
^{spaces} after~`|&|', hence you can conveniently finish a column by typing
`|&|'~at the end of a line in your input file, without worrying that an
extra blank space will be introduced there. \ (The second-last line of the
example ends with~`|&|', and there is an implicit blank space following that
symbol; if \TeX\ hadn't ignored that space, the words `the fourth'
wouldn't have started exactly at the beginning of the fourth column.) \
Incidentally, plain \TeX\ also ignores spaces after `|\+|', so that the
first column is treated like the others. (5) The `^|\it|' in the last line
of the example causes only the first column to be italicized, even though
no ^{braces} were used to confine the range of italics, because \TeX\
implicitly inserts braces around each individual entry of an alignment.

\danger Once you have issued a |\settabs| command, the tabs remain set until you
reset them, even though you go ahead and type ordinary paragraphs as usual.
But if you enclose |\settabs| in |{...}|, the tabs defined inside a group
don't affect the tabs outside; `^|\global||\settabs|' will not do what
you might think it should.

\danger Tabbed lines usually are used between paragraphs, in the same
places where you would type ^|\line| or ^|\centerline| to get lines with
a special format. But it's also useful to put |\+|~lines inside a |\vbox|;
this makes it convenient to specify ^{displays} that contain aligned
material. For example, if you type
\begintt
$$\vbox{\settabs 3 \columns
  \+This is&a strange&example\cr
  \+of displayed&three-column&format.\cr}$$
\endtt
you get the following display:
$$\vbox{\settabs 3 \columns
  \+This is&a strange&example\cr
  \+of displayed&three-column&format.\cr}$$
In this case the first column doesn't appear flush left, because \TeX\
centers a box that is being displayed. Columns that end with |\cr| in
a |\+|~line are put into a box with their natural width; so the first
and second columns here are one-third of the |\hsize|, but the third column
is only as wide as the word `example'. We have used |$$| ^^{dollardollar}
in this construction even though no mathematics is involved, because |$$|
does other useful things; for example, it centers the box, and it inserts
space above and below.

People don't always want tabs to be equally spaced, so there's another
way to set them, by typing `|\+|\<sample line>|\cr|' immediately after
`|\settabs|'. In this case tabs are placed at the positions
of the |&|'s in the ^{sample line}, and the sample line itself does not appear
in the output. For example,
\begintt
\settabs\+\indent&Horizontal lists\quad&\cr % sample line
\+&Horizontal lists&Chapter 14\cr
\+&Vertical lists&Chapter 15\cr
\+&Math lists&Chapter 17\cr
\endtt
causes \TeX\ to typeset the following three lines of material:
\nobreak\medskip
\settabs\+\indent&Horizontal lists\quad&\cr
\+&Horizontal lists&Chapter 14\cr
\+&Vertical lists&Chapter 15\cr
\+&Math lists&Chapter 17\cr
\medbreak\noindent
The |\settabs| command in this example makes column~1 as wide as a paragraph
^^{indention, see indentation}
indentation; and column~2 is as wide as `Horizontal lists' plus one quad of
space. ^^|\quad| Only two tabs are set in this case, because only two |&|'s
appear in the sample line. \ (A sample line usually ends with~|&\cr|, as it
does here, because text material between the last tab and |\cr|
isn't used for anything.)

The first line of a table can't always be used as a sample line, because it
won't necessarily give the correct tab positions. In a large table you have
to look ahead and figure out the biggest entry in each column; the sample
line is then constructed by typing the widest first column, the widest
second column, etc., omitting the last column. Be sure to include some
extra space between columns in the sample line, so that the columns
won't touch each other.

\def\frac#1/#2{\leavevmode\kern.1em
  \raise.5ex\hbox{\the\scriptfont0 #1}\kern-.1em
  /\kern-.15em\lower.25ex\hbox{\the\scriptfont0 #2}}
\exercise Explain how to typeset the following table [from Beck,
Bertholle, and Child, {\sl Mastering the Art of French Cooking\/}
(New York: Knopf, 1961)]: % p283
^^{Beck, Simone} ^^{Bertholle, Louisette} ^^{Child, Julia}
\nobreak\medskip
\settabs\+\indent&10\frac1/2 lbs.\qquad&\it Servings\qquad&\cr
\+&\negthinspace\it Weight&\it Servings&
  {\it Approximate Cooking Time\/}*\cr
\smallskip
\+&8 lbs.&6&1 hour and 50 to 55 minutes\cr
\+&9 lbs.&7 to 8&About 2 hours\cr
\+&9\frac1/2 lbs.&8 to 9&2 hours and 10 to 15 minutes\cr
\+&10\frac1/2 lbs.&9 to 10&2 hours and 15 to 20 minutes\cr
\smallskip
\+&* For a stuffed goose,
  add 20 to 40 minutes to the times given.\cr
\answer Notice the uses of `|\smallskip|' here to separate the table heading
and footing from the table itself; such refinements are often worthwhile.
\begintt
\settabs\+\indent&10\frac1/2 lbs.\qquad&\it Servings\qquad&\cr
\+&\negthinspace\it Weight&\it Servings&
  {\it Approximate Cooking Time\/}*\cr
\smallskip
\+&8 lbs.&6&1 hour and 50 to 55 minutes\cr
\+&9 lbs.&7 to 8&About 2 hours\cr
\+&9\frac1/2 lbs.&8 to 9&2 hours and 10 to 15 minutes\cr
\+&10\frac1/2 lbs.&9 to 10&2 hours and 15 to 20 minutes\cr
\smallskip
\+&* For a stuffed goose,
  add 20 to 40 minutes to the times given.\cr
\endtt
\def\frac#1/#2{\leavevmode\kern.1em
  \raise.5ex\hbox{\the\scriptfont0 #1}\kern-.1em
  /\kern-.15em\lower.25ex\hbox{\the\scriptfont0 #2}}%
The title line specifies `|\it|' three times, because each entry between
tabs is treated as a group by \TeX; you would get error messages galore
if you tried to say something like \hbox{`|\+&{\it Weight&Servings&...}\cr|'}.
The `^|\negthinspace|' in the title line is a small backspace that
compensates for the slant in an italic {\it W\/}; the author inserted
this somewhat unusual correction after seeing how the table looked
without it, on the first proofs. \ (You weren't supposed to think of this,
but it has to be mentioned.) \ See exercise 11.\fracexno\ for the `|\frac|'
macro; it's better to say `\frac1/2' than `$1\over2$', in a cookbook.\par
Another way to treat this table would be to display it in a vbox, instead
of including a first column whose sole purpose is to specify indentation.

\ninepoint % it's all dangerous from here to end of chapter
\danger If you want to put something ^{flush right} in its column, just type
`^|\hfill|' before it; and be sure to type `|&|' after it, so that
\TeX\ will be sure to move the information all the way until it touches
the next tab. Similarly, if you want to ^{center} something in its
column, type `|\hfill|' before it and `|\hfill&|' after it. For example,
\begintt
\settabs 2 \columns
\+\hfill This material is set flush right&
    \hfill This material is centered\hfill&\cr
\+\hfill in the first half of the line.&
    \hfill in the second half of the line.\hfill&\cr
\endtt
produces the following little table:\enddanger
\nobreak\medskip
\settabs 2 \columns
\+\hfill This material is set flush right&
    \hfill This material is centered\hfill&\cr
\+\hfill in the first half of the line.&
    \hfill in the second half of the line.\hfill&\cr

\danger The |\+| macro in Appendix~B works
by putting the \<text> for each column that's followed by~|&|
into an hbox as follows:
\begindisplay
|\hbox to |\<column width>|{|\<text>|\hss}|
\enddisplay
The ^|\hss| means that the text is normally flush left, and that it can
extend to the right of its box. Since |\hfill| is ``more infinite'' than
|\hss| in its ability to stretch, it has the effect of right-justifying or
centering as stated above. Note that |\hfill| doesn't shrink, but |\hss|
does; if the text doesn't fit in its column, it will stick out at the right.
You could avoid this by adding |\hskip| |0pt| |minus-1fil|; then
an oversize text would produce an overfull box.
You could also center some text by putting `|\hss|' before it and just
`|&|' after it; in that case the text would be allowed to extend to the
left and right of its column.
 The last column of a |\+|~line (i.e., the column entry that is
followed by |\cr|) is treated differently: The
\<text> is simply put into an hbox with its natural~width.\looseness=-1

\danger ^{Computer programs} present difficulties of a different kind, since
some people like to adopt a style in which the tab positions change from
line to line. For example, consider the following program fragment:
$$\vbox{\+\bf if $n<r$ \cleartabs&\bf then $n:=n+1$\cr
  \+&\bf else &{\bf begin} ${\it print\_totals}$; $n:=0$;\cr
  \+&&{\bf end};\cr
  \+\bf while $p>0$ do\cr
  \+\quad\cleartabs&{\bf begin} $q:={\it link}(p)$;
    ${\it free\_node}(p)$; $p:=q$;\cr
  \+&{\bf end};\cr}$$
Special tabs have been set up so that `{\bf then}' and `{\bf else}' appear
one above the other, and so do `{\bf begin}' and `{\bf end}'. It's possible
to achieve this by setting up a new sample line whenever a new tab position
is needed; but that's a tedious job, so plain \TeX\ makes it a little simpler.
Whenever you type |&| to the right of all existing tabs, the effect is to
set a new tab there, in such a way that the column just completed will have
its natural width. Furthermore, there's an operation `^|\cleartabs|' that
resets all tab positions to the right of the current column. Therefore the
computer program above can be \TeX ified as follows:
\begindisplay
|$$\vbox{\+\bf if $n<r$ \cleartabs&\bf then $n:=n+1$\cr|\cr
|  \+&\bf else &{\bf begin} ${\it print\_totals}$; $n:=0$;\cr|\cr
|  \+&&{\bf end};\cr|\cr
|  |\<The remaining part is left as an exercise>|}$$|\cr
\enddisplay

\dangerexercise Complete the example computer program by specifying three more
|\+|~lines.
\answer In such programs it seems best to type |\cleartabs| just before |&|,
whenever it is desirable to reset the old tabs. Multiletter identifiers look
best when set in ^{text italics} with ^|\it|, as explained in Chapter~18.
Thus, the following is recommended:
\begintt
\+\bf while $p>0$ do\cr
  \+\quad\cleartabs&{\bf begin} $q:={\it link}(p)$;
    ${\it free\_node}(p)$; $p:=q$;\cr
  \+&{\bf end};\cr  % note that the semicolon isn't bold
\endtt

\danger Although |\+| lines can be used in vertical boxes, you must never
use |\+| inside of another |\+| line. The |\+| macro is intended for
simple applications only.

\ddanger The |\+| and |\settabs| macros of Appendix B keep track of tabs by
maintaining register |\box|^|\tabs| as a box full of empty boxes whose
widths are the column widths in reverse order. Thus you can examine the
tabs that are currently set, by saying `^|\showbox||\tabs|'; this puts
the column widths into your log file, from right to left. For example,
after `|\settabs\+\hskip100pt&\hskip200pt&\cr\showbox\tabs|', \TeX\
will show the lines
\begintt
\hbox(0.0+0.0)x300.0
.\hbox(0.0+0.0)x200.0
.\hbox(0.0+0.0)x100.0
\endtt

\ddangerexercise Study the |\+| macro in Appendix B and figure out how to
change it so that tabs work as they do on a mechanical typewriter
(i.e., so that `|&|' always moves to the next tab that lies strictly
to the right of the current position). Assume that the user
doesn't backspace past previous tab positions; for example, if the input is
\hbox{`|\+&&\hskip-2em&x\cr|'}, do not bother to put `x' in the first or
second column, just put it at the beginning of the third column. \
(This exercise is a bit difficult.)
\answer Here we retain the idea that |&| inserts a new tab, when there
are no tabs to the right of the current position. Only one of the macros
that are used to process |\+|~lines needs to be changed; but
(unfortunately) it's the most complex one:
\begintt
\def\t@bb@x{\if@cr\egroup % now \box0 holds the column
  \else\hss\egroup \dimen@=0\p@
    \dimen@ii=\wd0 \advance\dimen@ii by1sp
    \loop\ifdim \dimen@<\dimen@ii
      \global\setbox\tabsyet=\hbox{\unhbox\tabsyet
        \global\setbox1=\lastbox}%
      \ifvoid1 \advance\dimen@ii by-\dimen@
        \advance\dimen@ii by-1sp \global\setbox1
          =\hbox to\dimen@ii{}\dimen@ii=-1pt\fi
      \advance\dimen@ by\wd1 \global\setbox\tabsdone
        =\hbox{\box1\unhbox\tabsdone}\repeat
    \setbox0=\hbox to\dimen@{\unhbox0}\fi
  \box0}
\endtt

\danger \TeX\ has another important way to make tables, using an operation
called ^|\halign| (``horizontal alignment''). In this case the table format
is based on the notion of a {\sl^{template}}, not on tabbing; the idea
is to specify a separate environment for the text in each column.
Individual entries are inserted into their templates, and presto, the
table is complete.

\danger For example, let's go back to the Horizontal/Vertical/Math list
example that appeared earlier in this chapter; we can specify it with
|\halign| instead of with tabs. The new specification is
\begintt
\halign{\indent#\hfil&\quad#\hfil\cr
  Horizontal lists&Chapter 14\cr
  Vertical lists&Chapter 15\cr
  Math lists&Chapter 17\cr}
\endtt
and it produces exactly the same result as the old one. This example
deserves careful study, because |\halign| is really quite simple once
you get the hang of it. The first line contains the {\sl ^{preamble}\/} to
the alignment, which is something like the sample line used to set tabs
for~|\+|. In this case the preamble contains two templates, namely
`|\indent#\hfil|' for the first column and `|\quad#\hfil|' for the
second. Each template contains exactly one appearance of `|#|', ^^{sharp}
and it means ``stick the text of each column entry in this place.''
Thus, the first column of the line that follows the preamble becomes
\begintt
\indent Horizontal lists\hfil
\endtt
when `|Horizontal lists|' is stuffed into its template; and the second
column, similarly, becomes `|\quad Chapter 14\hfil|'. The question is,
why |\hfil|? Ah, now we get to the interesting point of the whole thing:
\TeX\ reads an entire |\halign{...}| specification into its memory
before typesetting anything, and it keeps track of the maximum width
of each column, assuming that each column is set without stretching or
shrinking the glue. Then it goes back and puts every entry into a box,
setting the glue so that each box has the maximum column width. That's
where the |\hfil| comes in; it stretches to fill up the extra space in
narrower entries.

\dangerexercise What table would have resulted if the template for the
first column in this example had been `|\indent\hfil#|' instead of
`|\indent#\hfil|'?
\answer \par\nobreak\vskip-\baselineskip
\halign{\indent\hfil#&\quad#\hfil\cr
Horizontal lists&Chapter 14\cr\noalign{\nobreak}
Vertical lists&Chapter 15\cr\noalign{\nobreak}
Math lists&Chapter 17\qquad (i.e., the first column would be
  right-justified)\cr}

\danger Before reading further, please make sure that you understand the
idea of templates in the example just presented. There are several
important differences between |\halign| and~|\+|: (1)~|\halign| calculates
^^{halign compared to tabbing}
the maximum column widths automatically; you don't have to guess what the
longest entries will be, as you do when you set tabs with a sample line.
(2)~Each |\halign| does its own calculation of column widths; you have to
do something special if you want two different |\halign| operations to
produce identical alignments. By contrast, the |\+| operation remembers tab
positions until they are specifically reset; any number of paragraphs and
even |\halign| operations can intervene between |\+|'s, without affecting
the tabs. (3)~Because |\halign| reads an entire table in order to
determine the maximum column widths, it is unsuitable for huge tables
that fill several pages of a book. By contrast, the~|\+|~operation deals
with one line at a time, so it places no special demands on \TeX's memory.
\ (However, if you have a huge table, you should probably define your own
special-purpose macro for each line instead of relying on the general
|\+|~operation.) (4)~|\halign| takes less computer time than |\+|~does,
because |\halign| is a built-in command of \TeX, while |\+|~is a macro
that has been coded in terms of\/ |\halign| and various other primitive
operations. (5)~Templates are much more versatile than tabs, and they can
save you a lot of typing. For example, the Horizontal/Vertical/Math list
table could be specified more briefly by noticing that there's common
information in the columns:
\begintt
\halign{\indent# lists\hfil&\quad Chapter #\cr
  Horizontal&14\cr Vertical&15\cr Math&17\cr}
\endtt
You could even save two more keystrokes by noting that the chapter numbers
all start with `|1|'\thinspace! \ (Caution: It takes more time to think of
optimizations like this than to type things in a straightforward way;
do it only if you're bored and need something amusing to keep up
your interest.)\ (6)~On the other hand, templates are no substitute for
tabs when the tab positions are continually varying, as in the
computer program example.

\danger Let's do a more interesting table, to get more
experience with |\halign|. Here is another example based on the
^{Beck}/^{Bertholle}/^{Child} book cited earlier:
$$\vbox{\openup2pt
\halign{\hfil\bf#&\quad\hfil\it#\hfil&\quad\hfil#\hfil&
           \quad\hfil#\hfil&\quad#\hfil\cr
\sl American&\sl French&\sl Age&\sl Weight&\sl Cooking\cr
\noalign{\vskip-2pt}
\sl Chicken&\sl Connection&\sl(months)&\sl(lbs.)&\sl Methods\cr
\noalign{\smallskip}
Squab&Poussin&2&\frac3/4 to 1&Broil, Grill, Roast\cr
Broiler&Poulet Nouveau&2 to 3&1\frac1/2 to 2\frac1/2&Broil, Grill, Roast\cr
Fryer&Poulet Reine&3 to 5&2 to 3&Fry, Saut\'e, Roast\cr
Roaster&Poularde&5\frac1/2 to 9&Over 3&Roast, Poach, Fricassee\cr
Fowl&Poule de l'Ann\'ee&10 to 12&Over 3&Stew, Fricassee\cr
Rooster&Coq&Over 12&Over 3&Soup stock, Forcemeat\cr}}$$
Note that, except for the title lines, the first column is set right-justified
in boldface type; the middle columns are centered; the second column
is centered and in italics; the final column is left-justified. We would
like to be able to type the rows of the table as simply as possible; hence,
for example, it would be nice to be able to specify the bottom row by
typing only
\begintt
Rooster&Coq&Over 12&Over 3&Soup stock, Forcemeat\cr
\endtt
without worrying about type styles, centering, and so on. This not only
cuts down on keystrokes, it also reduces the chances for making typographical
errors. Therefore the template for the first column should be
`|\hfil\bf#|'; for the second column it should be `|\hfil\it#\hfil|' to
get the text centered and italicized; and so on. We also need to allow
for space between the columns, say one quad. {\it Voil\`a! La typographie
est sur la table:\/}\looseness=-1
\begindisplay
|\halign{\hfil\bf#&\quad\hfil\it#\hfil&\quad\hfil#\hfil&|\cr
|          \quad\hfil#\hfil&\quad#\hfil\cr|\cr
\ \<the title lines>\cr
| Squab&Poussin&2&\frac3/4 to 1&Broil, Grill, Roast\cr|\cr
| ... Forcemeat\cr}|\cr
\enddisplay
As with the |\+| operation, spaces are ignored after |&|, in the preamble
as well as in the individual rows of the table. Thus, it is convenient
to end a long row with `|&|' when the~row takes up more than one line
in your input file.

\dangerexercise How was the `{\bf Fowl}' line typed? \ (This is too easy.)
\answer |Fowl&Poule de l'Ann\'ee&10 to 12&Over 3&Stew, Fricassee\cr|

\danger The only remaining problem in this example is to specify the title
lines, which have a different format from the others. In this case the style
is different only because the typeface is slanted, so there's no special
difficulty; we just type
\begintt
\sl American&\sl French&\sl Age&\sl Weight&\sl Cooking\cr
\sl Chicken&\sl Connection&\sl(months)&\sl(lbs.)&\sl Methods\cr
\endtt
It is necessary to say `|\sl|' each time, because each individual entry
of a table is implicitly enclosed in braces.

\danger The author used `^|\openup||2pt|' to increase the distance between
baselines in the ^{poultry} table; a discriminating reader will notice
that there's also a bit of extra space between the title line and the
other lines.  This extra space was inserted by typing
`^|\noalign||{\smallskip}|' just after the title line. In general, you can say
\begindisplay
|\noalign{|\<vertical mode material>|}|
\enddisplay
just after any |\cr| in an |\halign|; \TeX\ will simply copy the vertical
mode material, without subjecting it to alignment, and it will appear
in place when the |\halign| is finished. You can use |\noalign| to
insert extra space, as here, or to insert penalties that affect page
breaking, or even to insert lines of text (see Chapter~19). Definitions
inside the braces of\/ |\noalign{...}| are local to that group.

\danger The |\halign| command also makes it possible for you to adjust
the spacing between columns so that a table will fill a specified area.
You don't have to decide that the ^{inter-column space} is a quad; you can
let \TeX\ make the decisions, based on how wide the columns come out,
because \TeX\ puts ``^{tabskip glue}'' between columns. This tabskip glue
is usually zero, but you can set it to any value you like by saying
`^|\tabskip||=|\<glue>'. For example,
let's do the poultry table again, but with the beginning of the
specification changed as follows:
\begintt
\tabskip=1em plus2em minus.5em
\halign to\hsize{\hfil\bf#&\hfil\it#\hfil&\hfil#\hfil&
           \hfil#\hfil&#\hfil\cr
\endtt
The main body of the table is unchanged, but the |\quad| spaces have been
removed from the preamble, and a nonzero |\tabskip| has been specified
instead. Furthermore `|\halign|' has been changed to `|\halign
to\hsize|'; this means that each line of the table will be put into a
box whose width is the current value of\/ ^|\hsize|, i.e., the horizontal
line width usually used in paragraphs. The resulting table looks like this:
$$\vbox{\openup2pt
\tabskip=1em plus2em minus.5em
\halign to\hsize{\hfil\bf#&\hfil\it#\hfil&\hfil#\hfil&
           \hfil#\hfil&#\hfil\cr
\sl American&\sl French&\sl Age&\sl Weight&\sl Cooking\cr
\noalign{\vskip-2pt}
\sl Chicken&\sl Connection&\sl(months)&\sl(lbs.)&\sl Methods\cr
\noalign{\smallskip}
Squab&Poussin&2&\frac3/4 to 1&Broil, Grill, Roast\cr
Broiler&Poulet Nouveau&2 to 3&1\frac1/2 to 2\frac1/2&Broil, Grill, Roast\cr
Fryer&Poulet Reine&3 to 5&2 to 3&Fry, Saut\'e, Roast\cr
Roaster&Poularde&5\frac1/2 to 9&Over 3&Roast, Poach, Fricassee\cr
Fowl&Poule de l'Ann\'ee&10 to 12&Over 3&Stew, Fricassee\cr
Rooster&Coq&Over 12&Over 3&Soup stock, Forcemeat\cr}}$$

\danger In general, \TeX\ puts tabskip glue before the first column, after
the last column, and between the columns of an alignment. You can specify
the final aligned size by saying `|\halign to|\<dimen>' or
`|\halign spread|\<dimen>', ^^|to| ^^|spread|
just as you can say `|\hbox to|\<dimen>' and `|\hbox spread|\<dimen>'.
This specification governs the setting of the tabskip glue; but it does
not affect the setting of the glue within column entries. \ (Those
entries have already been packaged into boxes having the maximum
natural width for their columns, as described earlier.)

\ddanger Therefore `|\halign| |to| |\hsize|' will do nothing if the
tabskip glue has no stretchability or shrinkability, except that it will
cause \TeX\ to report an ^{underfull} or ^{overfull} box. An overfull box
occurs if the tabskip glue can't shrink to meet the
given specification; in this case you get a warning on the terminal
and in your log file, but there is no ``^{overfull rule}'' to mark the
oversize table on the printed output. The warning message shows a
``^{prototype row}'' (see Chapter~27).

\danger The poultry example just given used the same tabskip glue
everywhere, but you can vary it by resetting ^|\tabskip| within the
preamble. The tabskip glue that is in force when \TeX\ reads the
`|{|' following |\halign| will be used before the first column;
the tabskip glue that is in force when \TeX\ reads the `|&|' after
the first template will be used between the first and second
columns; and so on. The tabskip glue that is in force when \TeX\
reads the |\cr| after the last template will be used after the
last column. For example, in
\begintt
\tabskip=3pt
\halign{\hfil#\tabskip=4pt& #\hfil&
  \hbox to 10em{\hss\tabskip=5pt # \hss}\cr ...}
\endtt
the preamble specifies aligned lines that will consist of the following
seven parts:
\begindisplay
tabskip glue $3\pt$;\cr
first column, with template `|\hfil#|';\cr
tabskip glue $4\pt$;\cr
second column, with template `|#\hfil|';\cr
tabskip glue $4\pt$;\cr
third column, with template `|\hbox to 10em{\hss# \hss}|';\cr
tabskip glue $5\pt$.\cr
\enddisplay

\ddanger \TeX\ copies the templates without interpreting them except to
remove any |\tabskip| glue specifications. More precisely, the tokens of the
preamble are passed directly to the templates without macro expansion;
\TeX\ looks only for `|\cr|' commands, `|&|', `|#|', `|\span|', and
`|\tabskip|'. The \<glue> following `|\tabskip|' is scanned in~the usual
way (with macro expansion), and the corresponding tokens are not included~in
the current template.  Notice that, in the example above, the space
after `|5pt|' also disappeared. The fact that |\tabskip=5pt| occurred
inside an extra level of braces did not make the definition local, since
\TeX\ didn't ``see'' those braces; similarly, if\/ |\tabskip| had been
preceded by `|\global|', \TeX\ wouldn't have made a global definition, it
would just have put `|\global|' into the template.  All assignments to
|\tabskip| within the preamble are local to the |\halign| (unless
^|\globaldefs| is positive), so the value of\/ |\tabskip| will be $3\pt$ again
when this particular |\halign| is completed.

\ddanger When `^|\span|' appears in a preamble, it causes the next token
to be expanded, i.e., ``ex-span-ded,'' before \TeX\ reads on.

\def\\{{\it c\/}}
\dangerexercise Design a preamble for the following table:
$$\halign to\hsize{\sl#\hfil\tabskip=.5em plus.5em&
   #\hfil\tabskip=0pt plus.5em&
   \hfil#\tabskip=1em plus2em&
  \sl#\hfil\tabskip=.5em plus.5em&
   #\hfil\tabskip=0pt plus.5em&
   \hfil#\tabskip=0pt\cr
England&P. Philips&1560--1628&
  Netherlands&J. P. Sweelinck&1562--1621\cr
&J. Bull&\\1563--1628&
  &P. Cornet&\\1570--1633\cr
Germany&H. L. Hassler&1562--1612&
  Italy&G. Frescobaldi&1583--1643\cr
&M. Pr\ae torius&1571--1621&
  Spain&F. Correa de Arauxo&\\1576--1654\cr
France&J. Titelouze&1563--1633&
  Portugal&M. R. Coelho&\\1555--\\1635\cr}$$
The tabskip glue should be zero at the left and right of each line; it should be
$1\em$ plus $2\em$ in the center; and it should be $.5\em$
plus $.5\em$ before the names, $0\em$ plus $.5\em$ before
the dates. Assume that the lines of the table will be specified by, e.g.,
\begintt
France&J. Titelouze&1563--1633&
  Portugal&M. R. Coelho&\\1555--\\1635\cr
\endtt
where `|\\|' has been predefined by `|\def\\{{\it c\/}}|'.
^^{organists}
^^{Cornet, Peeter} ^^{Philips, Peter} ^^{Sweelinck, Jan Pieterszoon}
^^{Bull, John} ^^{Titelouze, Jehan} ^^{Hassler, Hans Leo}
^^{Pr\ae torius [Schultheiss], Michael} ^^{Frescobaldi, Girolamo}
^^{Coelho, Manuel Rodrigues} % so listed in Lisbon & Rio, contrary to Groves!
^^{Correa de Arauxo, Francisco}
% The idea for this table came from The Organ and its Music, by Peeters and
% Vente (Antwerp, 1971); but their data was so flaky, I'm not citing them...
\answer |$$\halign to\hsize{\sl#\hfil\tabskip=.5em plus.5em&|\parbreak
        |   #\hfil\tabskip=0pt plus.5em&|\parbreak
        |   \hfil#\tabskip=1em plus2em&|\parbreak
        |  \sl#\hfil\tabskip=.5em plus.5em&|\parbreak
        |   #\hfil\tabskip=0pt plus.5em&|\parbreak
        |   \hfil#\tabskip=0pt\cr ...}$$|

\medskip
\ddangerexercise Design a preamble so that the table ^^{Welsh conjugation}
$$\def\welshverb#1={{\bf#1} = }
\halign to\hsize{\welshverb#\hfil\tabskip=1em plus1em&
  \welshverb#\hfil&\welshverb#\hfil\tabskip=0pt\cr
rydw i=I am&ydw i=am I&roeddwn i=I was\cr
rwyt ti=thou art&wyt ti=art thou&roeddet ti=thou wast\cr
mae e=he is&ydy e=is he&roedd e=he was\cr
mae hi=she is&ydy hi=is she&roedd hi=she was\cr
rydyn ni=we are&ydyn ni=are we&roedden ni=we were\cr
rydych chi=you are&ydych chi=are you&roeddech chi=you were\cr
maen nhw=they are&ydyn nhw=are they&roedden nhw=they were\cr}$$
can be specified by typing lines like
\begintt
mae hi=she is&ydy hi=is she&roedd hi=she was\cr
\endtt
\answer The trick is to define a new macro for the preamble:
\begintt
$$\def\welshverb#1={{\bf#1} = }
\halign to\hsize{\welshverb#\hfil\tabskip=1em plus1em&
  \welshverb#\hfil&\welshverb#\hfil\tabskip=0pt\cr ...}$$
\endtt

\setbox0=\vbox{\lineskip0pt
\tabskip=0pt plus1fil\halign to\hsize{\tabskip=0pt\strut
\hfil#: &\vtop{\parindent=0pt\hsize=16em\hangindent.5em\strut#\strut}\cr
\omit\hfil\sevenrm B.C.&\cr
397&War between Syracuse and Carthage\cr
396&Aristippus of Cyrene and An\-tis\-the\-nes of Athens (philosophers)\cr
395&Athens rebuilds the Long Walls\cr
394&Battles of Coronea and Cnidus\cr
\\393&Plato's
  {\sl Apology\/};
 Xenophon's
  {\sl Memorabilia\/};
 Aristophanes'
  {\sl Ecclesiazus\ae\/}\cr
391--87&Dionysius subjugates south Italy\cr
391&Isocrates opens his school\cr
390&Evagoras Hellenizes Cyprus\cr
387&``King's Peace''; Plato visits Ar\-chy\-tas of Taras (mathematician)
  and Dionysius I\cr
386&Plato founds the Academy\cr
383&Spartans occupy Cadmeia at Thebes\cr
380&Isocrates' {\sl Panegyricus\/}\cr}}

\medskip
\ddangerexercise \hsize=13pc
The line breaks in the second column of the table at the right were chosen
by \TeX\ so that the second column was exactly 16~ems wide. Furthermore,
the author specified one of the rows of the table by typing
$$\halign{#\hfil\cr
|\\393&Plato's {\sl Apology\/};|\cr
|  Xenophon's|\cr
|   {\sl Memorabilia\/};|\cr
|  Aristophanes'|\cr
|   {\sl Ecclesiazus\ae\/}\cr|\cr}$$
Can you guess what preamble was used in the alignment?  \ [The data comes
from Will ^{Durant}'s {\sl The Life of Greece\/} (Simon \& Schuster, 1939).]
^^{Aristippus of Cyrene} ^^{Antisthenes of Athens}
^^{Plato} ^^{Xenophon} ^^{Aristophanes} ^^{Dionysius I of Syracuse}
^^{Isocrates} ^^{Evagoras of Salamis} ^^{Archytas of Taras}
\strut\vadjust{\vbox to 0pt{\vss\box0\kern0pt}}% insert the aligned table
\answer |\hfil#: &\vtop{\parindent=0pt\hsize=16em|\parbreak
        |    \hangindent.5em\strut#\strut}\cr|\par\nobreak\medskip\noindent
With such narrow measure and such long words, the ^|\tolerance| should probably
also have been increased to, say, 1000 inside the ^|\vtop|; luckily it turned
^^|\strut| out that a higher tolerance wasn't needed.\par
{\sl Note:\/} The stated preamble solves the problem and demonstrates
that \TeX's line-breaking capability can be used within tables. But this
particular table is not really a good example of the use of\/ |\halign|,
because \TeX\ could typeset it directly, using ^|\everypar| in an
appropriate manner to set up the hanging indentation, and using |\par|
instead of\/ |\cr|. For example, one could say
\begintt
\hsize20em \parindent0pt \clubpenalty10000 \widowpenalty10000
\def\history#1&{\hangindent4.5em
  \hbox to4em{\hss#1: }\ignorespaces}
\everypar={\history} \def\\{\leavevmode{\it c\/}}
\endtt
which spares \TeX\ all the work of\/ |\halign| but yields essentially the
same result. ^^|\leavevmode|

\danger Sometimes a template will apply perfectly to all but one or two of
the entries in a column. For example, in the exercise just given, the
colons in the first column of the alignment were supplied by the
template `|\hfil#:|\]'; but the very first entry in that column,
`{\sevenrm B.C.}', did not have a colon. \TeX\ allows you to escape from
the stated template in the following way: If the very first token
of an alignment entry is `^|\omit|' (after macro expansion), then
the template of the preamble is omitted; the trivial template `|#|'
is used instead. For example, `{\sevenrm B.C.}' was put into the table above
by typing `|\omit\hfil\sevenrm B.C.|' immediately after the preamble.
You can use |\omit| in any column, but it must come first; otherwise \TeX\
will insert the template that was defined in the preamble.

\ddanger If you think about what \TeX\ has to do when it's processing
|\halign|, you'll realize that the timing of certain actions is critical.
Macros are not expanded when the preamble is being read, except as
described earlier; but once the |\cr| at the end of the preamble has been
sensed, \TeX\ must look ahead to see if the next token is |\noalign| or
|\omit|, and macros are expanded until the next non-space token is found.
If the token doesn't turn out to be |\noalign| or |\omit|, it is put
back to be read again, and \TeX\ begins to read the template (still
expanding macros). The template has two parts, called the $u$ and~$v$ parts,
where $u$~precedes the~`|#|' and $v$~follows~it. When \TeX\ has finished
the $u$~part, its reading mechanism goes back to the token that was
neither |\noalign| nor |\omit|, and continues to read the entry until
getting to the |&| or~|\cr| that ends the entry; then the $v$~part of
the template is read. A special internal operation called ^|\endtemplate|
is always placed at the end of the $v$~part; this causes \TeX\ to put
the entry into an ``^{unset box}'' whose glue will be set later when
the final column width is known. Then \TeX\ is ready for another entry;
it looks ahead for |\omit| (and also for |\noalign|, after~|\cr|) and
the process continues in the same way.

\ddanger One consequence of the process just described is that it may be
dangerous to begin an entry of an alignment with |\if...|, ^^{conditionals}
or with any macro that will expand into a replacement text whose first token
is |\if...|; the reason is that the condition will be evaluated before the
template has been read. \ (\TeX\ is still looking to see whether an |\omit|
will occur, when the |\if| is being expanded.) \ For example, if\/ ^|\strut|
has been defined to be an abbreviation for
\begindisplay
^|\ifmmode|\<text for math modes>|\else|\<text for nonmath modes>|\fi|
\enddisplay
and if\/ |\strut| appears as the first token in some alignment entry,
then \TeX\ will expand it into the \<text for nonmath modes> even though
the template might be `|$#$|', because \TeX\ will not yet be in math
mode when it is looking for a possible |\omit|. Chaos will
probably ensue. Therefore the replacement text for |\strut| in
Appendix~B is actually
\begintt
\relax\ifmmode...
\endtt
and `|\relax|' has also been put into all other macros that might suffer
from such timing problems. Sometimes you do want \TeX\ to expand a
conditional before a template is inserted, but careful macro designers
watch out for cases where this could cause trouble.

\newdimen\digitwidth \setbox0=\hbox{\sixrm0} \digitwidth=\wd0
\danger When you're typesetting ^{numerical tables}, it's common practice
to line up the ^{decimal points} in a column. For example, if two numbers
like `0.2010' and `297.1' both appear in the same column, you're supposed
to produce `$\catcode`?=13 \def?{\kern\digitwidth}
??0.2010 \atop 297.1???$'. This result isn't especially pleasing to the
eye, but that's what people do, so you might have to conform to the practice.
One way to handle this is to treat the column as two columns, somewhat as
|\eqalign| treats one formula as two formulas; the `.'\ can be placed at
the beginning of the second half-column. But the author usually prefers to
use another, less sophisticated method, which takes advantage of the fact
that the digits 0,~1, \dots,~9 have the same width in most fonts: You can
choose a character that's not used elsewhere in the table, say `|?|',
and change it to an ^{active character} that produces a blank space exactly
equal to the width of a digit. Then it's usually no chore to put such nulls
into the table entries so that each column can be regarded as either
centered or right-justified or left-justified. For example, `|??0.2010|'
and `|297.1???|'\ have the same width, so their decimal points will line
up easily. Here is one way to set up `|?|'\ for this purpose:
\begintt
\newdimen\digitwidth
\setbox0=\hbox{\rm0}
\digitwidth=\wd0
\catcode`?=\active
\def?{\kern\digitwidth}
\endtt
The last two definitions should be local to some ^{group}, e.g., inside a
|\vbox|, so that `|?|'\ will resume its normal behavior when the table is
finished. ^^|\active|

\danger Let's look now at some applications to mathematics. Suppose first
that you want to typeset the small table
$$\vbox{\halign{$\hfil#$ =&&\ \hfil#\hfil\cr
n\phantom)&0&1&2&3&4&5&6&7&8&9&10&11&12&13&14&15&16&17&18&19&20&\dots\cr
{\cal G}(n)&1&2&4&3&6&7&8&16&18&25&32&11&64&31&128&10&256&5&512&28&
1024&\dots\cr}}$$ % The Grundy function for SYM [cf. Winning Ways p441]
as a ^{display}ed equation. A brute force approach using |\eqalign| or
|\atop| is cumbersome because ${\cal G}(n)$ and $n$ don't always have the
same number of digits. It would be much nicer to type
\begindisplay
|$$\vbox{\halign{|\<preamble>|\cr|\cr
|     n\phantom)&0&1&2&3& ... &20&\dots\cr|\cr
|     {\cal G}(n)&1&2&4&3& ... &1024&\dots\cr}}$$|\cr
\enddisplay
for some \<preamble>. On the other hand, the \<preamble> is sure to be
long, since this table has 23 columns; so it looks as though |\settabs|
and |\+| will be easier. \TeX\ has a handy feature that helps a lot
in cases like this: Preambles often have a periodic structure,
^^{periodic preambles} ^^{cyclic preambles} ^^{ampersand ampersand}
and if you put an extra `|&|' ^^{ampersand} just before one of the templates,
\TeX\ will consider the preamble to be an infinite sequence that begins
again at the marked template when the |\cr| is reached. For example,
\begindisplay
$t_1\,$|&|$\,t_2\,$|&|$\,t_3\,$|&&|$\,t_4\,$|&|$\,t_5\,$|\cr|
\ is treated like \
$t_1\,$|&|$\,t_2\,$|&|$\,t_3\,$|&|$\,t_4\,$|&|$\,t_5\,$%
  |&|$\,t_4\,$|&|$\,t_5\,$|&|$\,t_4\,$|&|$\,\,\cdots\,$\cr
\noalign{\hbox{and}}
|&|$t_1\,$|&|$\,t_2\,$|&|$\,t_3\,$|&|$\,t_4\,$|&|$\,t_5\,$|\cr|
\ is treated like \
$t_1\,$|&|$\,t_2\,$|&|$\,t_3\,$|&|$\,t_4\,$|&|$\,t_5\,$%
  |&|$\,t_1\,$|&|$\,t_2\,$|&|$\,t_3\,$|&|$\,\,\cdots\,.$\cr
\enddisplay
The tabskip glue following each template is copied with that template.
The preamble will grow as long as needed, based on the number of columns
actually used by the subsequent alignment entries. Therefore all it takes is
\begintt
$\hfil#$ =&&\ \hfil#\hfil\cr
\endtt
to make a suitable \<preamble> for the ${\cal G}(n)$ problem.

\ddanger Now suppose that the task is to typeset three pairs of displayed
formulas, with all of the =~signs lined up: % cf. ACP Section 3.3.4
$$\vcenter{\openup1\jot \halign{
$\hfil#$&&${}#\hfil$&\qquad$\hfil#$\cr
V_i&=v_i-q_iv_j,&X_i&=x_i-q_ix_j,&
  U_i&=u_i,\qquad\hbox{for $i\ne j$};\cr
V_j&=v_j,&X_j&=x_j,&
  U_j&=u_j+\sum_{i\ne j}q_iu_i.\cr}}\eqno(23)$$
It's not easy to do this with three ^|\eqalign|'s, because the $\sum$ with
a subscript `$i\ne j$' makes the right-hand pair of formulas bigger than the
others; the baselines won't agree unless ``^{phantoms}'' are put into the
other two |\eqalign|'s (see Chapter~19). Instead of using |\eqalign|,
which is defined in Appendix~B to be a macro that uses |\halign|, let's
try to use |\halign| directly. The natural way to approach this display is
^^|\jot| to type
\begindisplay
|$$\vcenter{\openup1\jot \halign{|\<preamble>|\cr|\cr
|   |\<first line>|\cr |\<second line>|\cr}}\eqno(23)$$|\cr
\enddisplay
because the ^|\vcenter| puts the lines into a box that is properly centered
with respect to the equation number `(23)'; the ^|\openup| macro puts a
bit of extra space between the lines, as mentioned in Chapter~19.

\ddanger OK, now let's figure out how to type the \<first line> and \<second
line>. The usual convention is to put `|&|' before the symbols that we want
to line up, so the obvious solution is to type
\begintt
V_i&=v_i-q_iv_j,&X_i&=x_i-q_ix_j,&
  U_i&=u_i,\qquad\hbox{for $i\ne j$};\cr
V_j&=v_j,&X_j&=x_j,&
  U_j&=u_j+\sum_{i\ne j}q_iu_i.\cr
\endtt
Thus the alignment has six columns. We could take common elements into
the preamble (e.g., `|V_|' and `|=v_|'), but that would be too error-prone
and too tricky.

\ddanger The remaining problem is to construct a preamble to
support those lines. To the left of the =~signs we want the column to
be filled at the left; to the right of the =~signs we want it to be filled
at the right. There's a slight complication because we are breaking a math
formula into two separate pieces, yet we want the result to have the same
spacing as if it were one formula. Since we're putting the `|&|' just before
a relation, the solution is to insert `|{}|' ^^{lbrace rbrace}
at the beginning of the right-hand formula; \TeX\ will put the proper
space before the equals sign in `|${}=...$|', but it puts no space before
the equals sign in `|$=...$|'.  Therefore the desired \<preamble> is
\begintt
$\hfil#$&${}#\hfil$&
  \qquad$\hfil#$&${}#\hfil$&
  \qquad$\hfil#$&${}#\hfil$
\endtt
The third and fourth columns are like the first and second, except for
the |\qquad| that separates the equations; the fifth and sixth columns
are like the third and fourth. Once again we can use the handy `|&&|'
shortcut ^^{ampersand ampersand} to reduce the preamble to
\begintt
$\hfil#$&&${}#\hfil$&\qquad$\hfil#$
\endtt
With a little practice you'll find that it becomes easy to compose
preambles as you are typing a manuscript that needs them. However, most
manuscripts don't need them, so it may be a~while before you acquire even
a little practice in this regard.

\ddangerexercise Explain how to produce the following display:
$$\openup1\jot \tabskip=0pt plus1fil
\halign to\displaywidth{\tabskip=0pt
  $\hfil#$&$\hfil{}#{}$&
  $\hfil#$&$\hfil{}#{}$&
  $\hfil#$&$\hfil{}#{}$&
  $\hfil#$&${}#\hfil$\tabskip=0pt plus1fil&
  \llap{#}\tabskip=0pt\cr
10w&+&3x&+&3y&+&18z&=1,&(9)\cr
6w&-&17x&&&-&5z&=2.&(10)\cr}$$ % cf. ACP Eqs. 4.5.2-17,18
\answer The equation is divided into separate parts for terms and
plus/minus signs, and tabskip glue is used to center it:
\begintt
$$\openup1\jot \tabskip=0pt plus1fil
\halign to\displaywidth{\tabskip=0pt
  $\hfil#$&$\hfil{}#{}$&
  $\hfil#$&$\hfil{}#{}$&
  $\hfil#$&$\hfil{}#{}$&
  $\hfil#$&${}#\hfil$\tabskip=0pt plus1fil&
  \llap{#}\tabskip=0pt\cr
10w&+&3x&+&3y&+&18z&=1,&(9)\cr
6w&-&17x&&&-&5z&=2.&(10)\cr}$$
\endtt

\ddanger The next level of complexity occurs when some entries of a table
span two or more columns. \TeX\ provides two ways to handle this. First
^^{spanned columns in tables}
there's ^|\hidewidth|, which plain \TeX\ defines to be equivalent to
\begintt
\hskip-1000pt plus 1fill
\endtt
In other words, |\hidewidth| has an extremely negative ``natural width,''
but it will stretch without limit. If you put |\hidewidth| at the right of
some entry in an alignment, the effect is to ignore the width of this
entry and to let it stick out to the right of its box. \ (Think about it;
this entry won't be the widest one, when |\halign| figures the column
width.) \ Similarly, if you put |\hidewidth| at the left of an entry, it will
stick out to the left; and you can put |\hidewidth| at both left and right,
as we'll see later.

\ddanger The second way to handle table entries that span columns is to use
the ^|\span| primitive, which can be used instead of `|&|' in any
line of the table. \ (We've already seen that |\span| means ``expand'' in
preambles; but outside of preambles its use is
completely different.) \ When `|\span|' appears in place of `|&|',
the material before and after the |\span| is processed in the ordinary
way, but afterward it is placed into a single box instead of two boxes.
The width of this combination box is the sum of the individual column
widths plus the width of the tabskip glue between them; therefore the
spanning box will line up with non-spanning boxes in other rows.

\ddanger For example, suppose that there are three columns, with the
respective templates $u_1\,$|#|$\,v_1\,$|&| $u_2\,$|#|$\,v_2\,$|&|
$u_3\,$|#|$\,v_3$; suppose that the column widths are $w_1$, $w_2$,~$w_3$;
suppose that $g_0$,~$g_1$, $g_2$,~$g_3$ are the tabskip glue widths after
the glue has been set; and suppose that the line
\begindisplay
$a_1$|\span|$\,\,a_2$|\span|$\,\,a_3$|\cr|
\enddisplay
has appeared in the alignment. Then the material for
`$u_1a_1v_1u_2a_2v_2u_3a_3v_3$' (i.e., the result `$u_1a_1v_1$' of
column~1 followed by the results of columns 2 and~3) will be placed into
an hbox of width $w_1+g_1+ w_2+g_2+w_3$. That hbox will be preceded by
glue of width~$g_0$ and it will be followed by glue of width~$g_3$, in the
larger hbox that contains the entire aligned line.

\ddanger You can use ^|\omit| in conjunction with |\span|. For example,
if we continue with the notation of the previous paragraph, the line
\begindisplay
|\omit|$\,a_1\,$|\span|$\,a_2\,$|\span\omit|$\,a_3\,$|\cr|
\enddisplay
would put the material for `$a_1u_2a_2v_2a_3$' into the hbox just considered.

\ddanger It's fairly common to span several columns and to omit all their
templates, so plain \TeX\ provides a ^|\multispan| macro that spans
a given number of columns. For example, `|\multispan3|' expands into
`|\omit\span\omit\span\omit|'. If the number of spanned columns is
greater than~9, you must put it in braces, e.g., `|\multispan{13}|'.

\ddanger The preceding paragraphs are rather abstract, so let's look at
an example that shows what |\span| actually does. Suppose you type
\begintt
$$\tabskip=3em
\vbox{\halign{&\hrulefill#\hrulefill\cr
    first&second&third\cr
    first-and-second\span\omit&\cr
    &second-and-third\span\omit\cr
    first-second-third\span\omit\span\omit\cr}}$$
\endtt
The preamble specifies arbitrarily many templates equal to
`|\hrulefill#\hrulefill|'; the ^|\hrulefill| macro is like |\hfill|
except that the blank space is filled with a horizontal rule. Therefore
you can see the filling in the resulting alignment, which shows the
spanned columns:
$$\tabskip=3em
\vbox{\halign{&\hrulefill#\hrulefill\cr
    first&second&third\cr
    first-and-second\span\omit&\cr
    &second-and-third\span\omit\cr
    first-second-third\span\omit\span\omit\cr}}$$
The rules stop where the tabskip glue separates columns. You don't see
rules in the first line, since the entries in that line were the widest
in their columns. However, if the tabskip glue had been $1\em$ instead
of $3\em$, the table would have looked like this:
$$\tabskip=1em
\vbox{\halign{&\hrulefill#\hrulefill\cr
    first&second&third\cr
    first-and-second\span\omit&\cr
    &second-and-third\span\omit\cr
    first-second-third\span\omit\span\omit\cr}}$$

\ddangerexercise Consider the following table, which is called
^{Walter's worksheet}: ^^{IRS}
^^{Green, Walter} % from instructions to form 1040 (1982), p13
$$\halign{\indent
\hfil# &#\hfil&\quad#&\ \hfil#&\ \hfil#\cr
1&Adjusted gross income\dotfill\span\omit\span&\$4,000\cr
2&Zero bracket amount for&\cr
 &a single individual\dotfill\span\omit&\$2,300\cr
3&Earned income\dotfill\span\omit&\underbar{ 1,500}\cr
4&Subtract line 3 from line 2\dotfill\span\omit\span&\underbar{ 800}\cr
5&Add lines 1 and 4. Enter here\span\omit\span\cr
 &and on Form 1040, line 35\dotfill\span\omit\span&\$4,800\cr}
$$
Define a preamble so that the following specification will produce
Walter's worksheet.
$$\halign{\indent#\hfil\cr
|\halign{|\<preamble>|\cr|\cr
|  1&Adjusted gross income\dotfill\span\omit\span&\$4,000\cr|\cr
|  2&Zero bracket amount for&\cr|\cr
|   &a single individual\dotfill\span\omit&\$2,300\cr|\cr
|  3&Earned income\dotfill\span\omit&\underbar{ 1,500}\cr|\cr
|  4&Subtract line 3 from line 2\dotfill|\cr
|      \span\omit\span&\underbar{ 800}\cr|\cr
|  5&Add lines 1 and 4. Enter here\span\omit\span\cr|\cr
|   &and on Form 1040, line 35\dotfill\span\omit\span&\$4,800\cr}|\cr
}$$
(The macro ^|\dotfill| is like |\hrulefill| but it fills with dots;
the macro ^|\underbar| puts its argument into an hbox and underlines it.)
\answer |\hfil# &#\hfil&\quad#&\ \hfil#&\ \hfil#\cr|

\ddanger Notice the ``early'' appearance of\/ ^|\cr| in line~2 of the
previous exercise. You needn't have the same number of columns in every
line of an alignment; `|\cr|' means that there are no more columns
in the current line.

\medskip
\ddangerexercise Explain how to typeset the ^{generic matrix}
$\smash{\pmatrix{a_{11}&a_{12}&\ldots&a_{1n}\cr
             a_{21}&a_{22}&\ldots&a_{2n}\cr
             \multispan4\dotfill\cr
             a_{m1}&a_{m2}&\ldots&a_{mn}\cr}.}$
\answer |\pmatrix{a_{11}&a_{12}&\ldots&a_{1n}\cr|\parbreak
        |    a_{21}&a_{22}&\ldots&a_{2n}\cr|\parbreak
        |    \multispan4\dotfill\cr|\parbreak
        |    a_{m1}&a_{m2}&\ldots&a_{mn}\cr}|

\bigskip\medskip
\ddanger The presence of spanned columns adds a complication to \TeX's
rules for calculating column widths; instead of simply choosing the
maximum natural width of the column entries, it's also necessary to
make sure that the sum of certain widths is big enough to accommodate
spanned entries. So here is what \TeX\ actually does: First, if
any pair of adjacent columns is always spanned as a unit (i.e., if
there's a |\span| between them whenever either one is used), these
two columns are effectively merged into one and the tabskip glue between
them is set to zero. This reduces the problem to the case that
every tab position actually occurs at a boundary. Let there be $n$
columns remaining after such reductions, and for $1\le i\le j\le n$ let
$w_{ij}$ be the maximum natural width of all entries that span columns $i$
through~$j$, inclusive; if there are no such spanned entries, let
$w_{ij}=-\infty$. \ (The merging of dependent columns guarantees that, for
each~$j$, there exists $i\le j$ such that $w_{ij}>-\infty$.) \ Let $t_k$
be the natural width of the tabskip glue between columns $k$ and~$k+1$,
for $1\le k<n$. Now the final width $w_j$ of column~$j$ is determined by
the formula
\begindisplay
$\displaystyle w_j=\max_{1\le i\le j}\textstyle\bigl(w_{ij}
  -\sum_{i\le k<j}(w_k+t_k)\bigr)$
\enddisplay
for $j=1$, 2, \dots, $n$ (in this order). It follows that
$w_{ij}\le w_i+t_i+\cdots+t_{j-1}+w_j$, for all $i\le j$,
as desired.  After the widths~$w_j$ are determined, the tabskip amounts
may have to stretch or shrink; if they shrink, $w_{ij}$ might turn out to
be more than the final width of a box that spans columns $i$ through~$j$,
hence the glue in such a box might shrink.

\ddanger These formulas usually work fine, but sometimes they produce
undesirable effects. For example, suppose that $n=3$, $w_{11}=w_{22}=w_{33}
=10$, $w_{12}=w_{23}=-\infty$, and $w_{13}=100$; in other words, the columns
by themselves are quite narrow, but there's a big wide entry that's
supposed to span all three columns. In this case \TeX's formula makes
$w_1=w_2=10$ but $w_3=80-t_1-t_2$, so all the excess width is allocated
to the third column. If that's not what you want, the remedy is~to use
^|\hidewidth|, or to increase the natural width of the tabskip glue
between columns.

\ddanger The next level of complexity that occurs in tables is the
appearance of horizontal and vertical ruled lines. People who know
how to make ^{ruled tables} are generally known as \TeX\ Masters.
^^{TeX Masters} Are you ready?

\ddanger If you approach vertical rules in the wrong manner, they can be
difficult; but there {\sl is\/} a decent way to get them into tables
without shedding too many tears. The first step is to say
`^|\offinterlineskip|', which means that there will be no blank space
between lines; \TeX\ cannot be allowed to insert ^{interline glue} in its
normal clever way, because each line is supposed to contain a ^|\vrule|
that abuts another ^|\vrule| in the neighboring lines above and/or below.
We will put a strut into every line, by including one in the preamble;
then each line will have the proper height and depth, and there will be no
need for interline glue. \TeX\ puts every column entry of an alignment into an
hbox whose height and depth are set equal to the height and depth of the
entire line; therefore |\vrule| commands will extend to the top and bottom
of the lines even when their height and/or depth are not specified.

\ddanger A ``column'' should be allocated to every vertical rule, and such
a column can be assigned the template `|\vrule#|'.  Then you obtain a
vertical rule by simply leaving the column entries blank, in the normal
lines of the alignment; or you can say `|\omit|' if you want to omit the
rule in some line; or you can say `|height 10pt|' if you want a
nonstandard height; and so on.

\ddanger Here is a small table that illustrates the points just made.
\ [The data appeared in an article by A. H. ^{Westing}, {\sl BioScience\/
\bf31} (1981), 523--524.]
\def\BC{\hbox to2em{ \sc B.C.\hss}}%
\def\AD{\hbox to2em{ \sc A.D.\hss}}%
$$\hbox to\hsize{\vbox{\halign{\indent#\hfil\cr
|\vbox{\offinterlineskip|\cr
|\hrule|\cr
|\halign{&\vrule#&|\cr
|  \strut\quad\hfil#\quad\cr|\cr
|height2pt&\omit&&\omit&\cr|\cr
|&Year\hfil&&World Population&\cr|\cr
|height2pt&\omit&&\omit&\cr|\cr
|\noalign{\hrule}|\cr
|height2pt&\omit&&\omit&\cr|\cr
|&8000\BC&&5,000,000&\cr|\cr
|&50\AD&&200,000,000&\cr|\cr
|&1650\AD&&500,000,000&\cr|\cr
|&1850\AD&&1,000,000,000&\cr|\cr
|&1945\AD&&2,300,000,000&\cr|\cr
|&1980\AD&&4,400,000,000&\cr|\cr
|height2pt&\omit&&\omit&\cr}|\cr
|\hrule}|\cr
}}\hfill\vbox{\offinterlineskip
\halign{&\vrule#&
  \strut\quad\hfil#\quad\cr
\multispan5\hrulefill\cr
height2pt&\omit&&\omit&\cr
&Year\hfil&&World Population&\cr
height2pt&\omit&&\omit&\cr
\multispan5\hrulefill\cr
height2pt&\omit&&\omit&\cr
&8000\BC&&5,000,000&\cr
&50\AD&&200,000,000&\cr
&1650\AD&&500,000,000&\cr
&1850\AD&&1,000,000,000&\cr
&1945\AD&&2,300,000,000&\cr
&1980\AD&&4,400,000,000&\cr
height2pt&\omit&&\omit&\cr
\multispan5\hrulefill\cr}}}$$
In this example the first, third, and fifth columns are reserved for vertical
rules. Horizontal rules are obtained by saying `^|\hrule|' outside the
|\halign| or `^|\noalign||{\hrule}|' inside it, because the |\halign| appears
in a vbox whose width is the full table width. The horizontal
rules could also have been specified by saying `^|\multispan||5\hrulefill|'
inside the |\halign|, since that would produce a rule that spans all
five columns.

\ddanger The only other nonobvious thing about this table is the
inclusion of several lines that say
`|height2pt&\omit&&\omit&\cr|'; can you see what they do? The |\omit|
instructions mean that there's no numerical information, and they
also suppress the ^|\strut| from the line; the `|height2pt|' makes the
first |\vrule| $2\pt$ high, and the other two rules will follow suit.
Thus, the effect is to extend the vertical rules by two points, where
they touch the horizontal rules. This is a little touch that improves
the appearance of boxed tables; look for it as a mark of quality.

\ddangerexercise Explain why the lines of this table say `|&\cr|' instead of
just `|\cr|'.
\answer `|\cr|' would have omitted the final column, which is a vertical rule.

\ddanger Another way to get vertical rules into tables is to typeset without
them, then back up (using negative glue) and insert them.

\ddanger Here is another table; this one has become a classic, ever since
Michael ^{Lesk} published it as one of the first examples in his report
on a program to format tables [Bell Laboratories Computing Science
Technical Report {\bf 49} (1976)].  It illustrates several typical
problems that arise in connection with boxed information. In order to
demonstrate \TeX's ability to adapt a table to different circumstances,
tabskip glue is used here to adjust the column widths; the table appears
twice, once generated by `|\halign|~|to125pt|' and once by
`|\halign|~|to200pt|', with nothing else changed. ^^{AT\&T}
$$\hbox to\hsize{%
\vbox{\tabskip=0pt \offinterlineskip
\def\tablerule{\noalign{\hrule}}
\halign to125pt{\strut#&\vrule#\tabskip=1em plus2em&
  \hfil#&\vrule#&\hfil#\hfil&\vrule#&
  \hfil#&\vrule#\tabskip=0pt\cr\tablerule
&&\multispan5\hfil AT\&T Common Stock\hfil&\cr\tablerule
&&\omit\hidewidth Year\hidewidth&&
 \omit\hidewidth Price\hidewidth&&
 \omit\hidewidth Dividend\hidewidth&\cr\tablerule
&&1971&&41--54&&\$2.60&\cr\tablerule
&&   2&&41--54&&2.70&\cr\tablerule
&&   3&&46--55&&2.87&\cr\tablerule
&&   4&&40--53&&3.24&\cr\tablerule
&&   5&&45--52&&3.40&\cr\tablerule
&&   6&&51--59&&.95\rlap*&\cr\tablerule
\noalign{\smallskip}
&\multispan7* (first quarter only)\hfil\cr
}}\hfil
\vbox{\tabskip=0pt \offinterlineskip
\def\tablerule{\noalign{\hrule}}
\halign to200pt{\strut#&\vrule#\tabskip=1em plus2em&
  \hfil#&\vrule#&\hfil#\hfil&\vrule#&
  \hfil#&\vrule#\tabskip=0pt\cr\tablerule
&&\multispan5\hfil AT\&T Common Stock\hfil&\cr\tablerule
&&\omit\hidewidth Year\hidewidth&&
 \omit\hidewidth Price\hidewidth&&
 \omit\hidewidth Dividend\hidewidth&\cr\tablerule
&&1971&&41--54&&\$2.60&\cr\tablerule
&&   2&&41--54&&2.70&\cr\tablerule
&&   3&&46--55&&2.87&\cr\tablerule
&&   4&&40--53&&3.24&\cr\tablerule
&&   5&&45--52&&3.40&\cr\tablerule
&&   6&&51--59&&.95\rlap*&\cr\tablerule
\noalign{\smallskip}
&\multispan7* (first quarter only)\hfil\cr}}}$$
The following specification did the job:
\begindisplay
|\vbox{\tabskip=0pt \offinterlineskip|\cr
|\def\tablerule{\noalign{\hrule}}|\cr
|\halign to|\<dimen>|{\strut#& \vrule#\tabskip=1em plus2em&|\cr
|  \hfil#& \vrule#& \hfil#\hfil& \vrule#&|\cr
|  \hfil#& \vrule#\tabskip=0pt\cr\tablerule|\cr
|&&\multispan5\hfil AT\&T Common Stock\hfil&\cr\tablerule|\cr
|&&\omit\hidewidth Year\hidewidth&&|\cr
| \omit\hidewidth Price\hidewidth&&|\cr
| \omit\hidewidth Dividend\hidewidth&\cr\tablerule|\cr
|&&1971&&41--54&&\$2.60&\cr\tablerule|\cr
|&&   2&&41--54&&2.70&\cr\tablerule|\cr
|&&   3&&46--55&&2.87&\cr\tablerule|\cr
|&&   4&&40--53&&3.24&\cr\tablerule|\cr
|&&   5&&45--52&&3.40&\cr\tablerule|\cr
|&&   6&&51--59&&.95\rlap*&\cr\tablerule \noalign{\smallskip}|\cr
|&\multispan7* (first quarter only)\hfil\cr}}|\cr
\enddisplay
Points of interest are: (1)~The first column contains a strut; otherwise
it would have been necessary to put a strut on the lines that say
`AT\&T' and `(first quarter only)', since those lines omit the templates
of all other columns that might have a built-in strut. (2)~`^|\hidewidth|'
is used in the title line so that the width of columns will be affected
only by the width of the numeric data. (3)~`^|\rlap|' is used so that
the asterisk doesn't affect the alignment of the numbers.  (4)~If the
tabskip specification had been `|0em plus3em|' instead of `|1em plus2em|',
the alignment wouldn't have come out right, because `AT\&T Common Stock'
would have been wider than the natural width of everything it spanned; the
excess width would all have gone into the `Dividend' column.

\ddangerexercise Explain how to add $2\pt$ more space above and below
`AT\&T Common Stock'.
\answer One way is to include two lines just before and after the title
line, saying `|\omit&height2pt&\multispan5&\cr|'. Another way is to
put |\bigstrut| into some column of the title line, for some appropriate
invisible box |\bigstrut| of width zero. Either way makes the table
look better.

\ddangerexercise Typeset the following chart, making it exactly 
36\thinspace em wide:
^^{family tree}
^^{Bohning [Knuth], Louise Marie}
^^{Ehlert [Bohning], Pauline Anna Marie}
^^{B\"ohning, Martin John Henry}
^^{Wischmeyer [Ehlert], Clara Louise}
^^{Ehlert, Ernst Fred}
^^{Blase [B\"ohning], Maria Dorothea}
^^{B\"ohning, Jobst Heinrich}
$$\vbox{\tabskip=0pt \offinterlineskip
\halign to 36em{\tabskip=0pt plus1em#&
  #\hfil&#&#\hfil&#&#\hfil&#\tabskip=0pt\cr
&&&&&\strut J. H. B\"ohning, 1838&\cr
&&&&\multispan3\hrulefill\cr
&&&\strut M. J. H. B\"ohning, 1882&\vrule\cr
&&\multispan3\hrulefill\cr
&&\vrule&&\vrule&\strut M. D. Blase, 1840&\cr
&&\vrule&&\multispan3\hrulefill\cr
&\strut L. M. Bohning, 1912&\vrule\cr
\multispan3\hrulefill\cr
&&\vrule&&&\strut E. F. Ehlert, 1845&\cr
&&\vrule&&\multispan3\hrulefill\cr
&&\vrule&\strut P. A. M. Ehlert, 1884&\vrule\cr
&&\multispan3\hrulefill\cr
&&&&\vrule&\strut C. L. Wischmeyer, 1850&\cr
&&&&\multispan3\hrulefill\cr
}}$$
\answer The trick is to have ``empty'' columns at the extreme left and right;
then the |\hrulefill|'s are able to span the tabskip glue.
\begintt
$$\vbox{\tabskip=0pt \offinterlineskip
\halign to 36em{\tabskip=0pt plus1em#&
  #\hfil&#&#\hfil&#&#\hfil&#\tabskip=0pt\cr
&&&&&\strut J. H. B\"ohning, 1838&\cr
&&&&\multispan3\hrulefill\cr
&&&\strut M. J. H. B\"ohning, 1882&\vrule\cr
&&\multispan3\hrulefill\cr
&&\vrule&&\vrule&\strut M. D. Blase, 1840&\cr
&&\vrule&&\multispan3\hrulefill\cr
&\strut L. M. Bohning, 1912&\vrule\cr
\multispan3\hrulefill\cr
&&\vrule&&&\strut E. F. Ehlert, 1845&\cr
&&\vrule&&\multispan3\hrulefill\cr
&&\vrule&\strut P. A. M. Ehlert, 1884&\vrule\cr
&&\multispan3\hrulefill\cr
&&&&\vrule&\strut C. L. Wischmeyer, 1850&\cr
&&&&\multispan3\hrulefill\cr}}$$
\endtt

\ddanger If you're having trouble ^{debugging} an alignment, it sometimes
helps to put `^|\ddt|' at the beginning and end of the templates in your
preamble. This is an undefined control sequence that causes \TeX\ to stop,
displaying the rest of the template. When \TeX\ stops, you can use
|\showlists| and other commands to see what the machine thinks it's doing.
If \TeX\ doesn't stop, you know that it never reached that part of the
template.

\ddanger It's possible to have alignments within alignments. Therefore when
\TeX\ sees a `|&|' or `|\span|' or `|\cr|', it needs some way to decide which
alignment is involved. The rule is that an entry ends when `|&|' or `|\span|'
or `|\cr|' occurs at the same level of braces that was current when the
entry began; i.e., there must be an equal number of left and right ^{braces}
in every entry. For example, in the line
\begintt
\matrix{1&1\cr 0&1\cr}&\matrix{0&1\cr 0&0\cr}\cr
\endtt
\TeX\ will not resume the template for the first column when it is
scanning the argument to |\matrix|, because the |&|'s and |\cr|'s in
that argument are enclosed in braces. Similarly, |&|'s and |\cr|'s in
the preamble do not denote the end of a template unless the resulting
template would have an equal number of left and right braces.

\ddanger You have to be careful with the use of |&| and ^|\span| and ^|\cr|,
^^{ampersand}
because these tokens are intercepted by \TeX's scanner even when it is
not expanding macros. For example, if you say `|\let\x=\span|' in the
midst of an alignment entry, \TeX\ will think that the `|\span|' ends
the entry, so |\x| will become equal to the first token following the
`|#|' in the template. You can hide this |\span| by putting it in
braces; e.g., `|{\global\let\x=\span}|'. \ (And Appendix~D explains how to
avoid |\global| here.)

\ddanger Sometimes people forget the |\cr| on the last line of an
alignment. This can cause mysterious effects, because \TeX\ is not
clairvoyant. For example, consider the following apparently simple case:
\begintt
\halign{\centerline{#}\cr
  A centered line.\cr
  And another?}
\endtt
(Notice the missing |\cr|.) \ A curious thing happens here when \TeX\
processes the erroneous line, so please pay attention. The template
begins with `|\centerline{|', so \TeX\ starts to scan the argument to
|\centerline|. Since there's no `|\cr|' after the question mark, the `|}|'
after the question mark is treated as the end of the argument to
|\centerline|, {\sl not\/} as the end of the |\halign|. \TeX\ isn't going
to be able to finish the alignment unless the subsequent text has
the form `|...{...\cr|'. Indeed, an entry like `|a}b{c|' is legitimate
with respect to the template `|\centerline{#}|', since it yields
`|\centerline{a}b{c}|'; \TeX\ is correct when it gives no error message in
this case. But the computer's idea of the current situation is
different from the user's, so a puzzling error message will probably occur
a few lines later.

\ddanger To help avoid such situations, there's a primitive command
^|\crcr| that acts exactly like |\cr| except that it does nothing
when it immediately follows a |\cr| or a |\noalign{...}|. Thus, when
you write a macro like |\matrix|, you can safely insert |\crcr|
at the end of the user's argument; this will cover up an error if the
user forgot the final |\cr|, and it will cause no harm if the final
|\cr| was present.

\ddanger Are you tired of typing |\cr|? ^^{cr, avoiding}
You can get plain \TeX\ to insert an automatic |\cr| at the end of
each input line in the following way: ^^|\begingroup|
\begindisplay
|\begingroup \let\par=\cr \obeylines %|\cr
|\halign{|\<preamble>\cr
\ \ \ \<first line of alignment>\cr
\ \ \ \ \dots\cr
\ \ \ \<last line of alignment>\cr
|  }\endgroup|\cr
\enddisplay
This works because ^|\obeylines| makes the ASCII ^\<return> into
an active character that uses the current meaning of\/ ^|\par|, and
plain \TeX\ puts \<return> at the end of an input line (see Chapter~8).
If you don't want a~|\cr| at the end of a certain line,
just type `|%|' and the corresponding |\cr|
will be ``commented out.'' ^^{percent} \ (This special mode doesn't
work with ^|\+| lines, since |\+| is a macro whose argument is delimited
by the token `|\cr|', not simply by a token that has the same meaning
as~|\cr|.  ^^{delimited arguments} But you can redefine |\+| to overcome
this hurdle, if you want to. For example, define a macro |\alternateplus|
that is just like |\+| except that its argument is delimited by the active
character |^^M|; then include the command `|\let\+=\alternateplus|' as
part of\/ |\obeylines|.)

\danger The control sequence ^|\valign| is analogous to |\halign|, but
rows and columns change r\^oles. In this case |\cr| marks the bottom of
a column, and the aligned columns are vboxes that are put together in
horizontal mode. The individual entries of each column are vboxed with
depth zero (i.e., as if\/ ^|\boxmaxdepth| were zero, as explained in
Chapter~12); the entry heights for each row of a |\valign| are maximized
in the same fashion as the entry widths for each column of an~|\halign|
are maximized. The ^|\noalign| operation can now be used to insert
horizontal mode material between columns; the ^|\span| operation now
spans rows. ^^{spanned rows in tables} People usually work with \TeX\
at least a year before they find their first application for |\valign|;
and then it's usually a one-row `|\valign{\vfil#\vfil\cr...}|'.
But the general mechanism is there if you need it.

\endchapter

If sixteen pennies are arranged in the form of a square
there will be the same number of pennies in every row, every column,
and each of the two long diagonals.
Can you do the same with twenty pennies?
\author HENRY ERNEST ^{DUDENEY}, {\sl The Best Coin Problems\/} (1909)
  % Strand Magazine, July 1909, page 83; answer in August 1909, page 240
\immediate\write\ans{}
\immediate\write\ans{\string\ansno\chapno.$\infty$:}
\copytoblankline (Solution to Dudeney's problem.) \
Let |\one| and |\two| be macros that produce a vertical list
denoting one and two pennies, respectively. The problem can be
solved with ^|\valign| as follows:
\begintt
\valign{\vfil#&\vfil#&\vfil#&\vfil#\cr
  \two&\one&\one&\one\cr
  \one&\one&\two&\one\cr
  \one&\one&\one&\two\cr
  \one&\two&\one&\one\cr}
\endtt
Since |\valign| transposes rows and columns, the result is\quad
\setbox0=\hbox{\vbox{
    \def\pennytop{\hbox to 24pt{\manual\char'130\hfil}}%
      \def\pennyedge{\hbox{\manual\char'133}}%
      \def\one{\pennytop\pennyedge}%
      \def\two{\one\pennyedge}%
      \baselineskip0pt\lineskip0pt\tabskip=14pt
    \hbox{\valign{\vfil#&\vfil#&\vfil#&\vfil#\cr
        \two&\one&\one&\one\cr
        \one&\one&\two&\one\cr
        \one&\one&\one&\two\cr
        \one&\two&\one&\one\cr}}\kern-11pt}}%
\ht0=0pt \dp0=11pt \box0.

\bigskip

It was she who controlled the whole of the Fifth Column.
\author AGATHA ^{CHRISTIE}, {\sl N or M?\/} (1941) % chapter 5, part 1

\eject
\beginchapter Chapter 23. Output Routines

We investigated \TeX's page-building technique in Chapter 15, where we
discussed the basic two-stage strategy that is used: \TeX\ gathers
material until it has accumulated more than will fit on a page; then it
spews out one page of data, based on what it thinks is the best breakpoint
between pages; then it returns to gather material for the next page in
the same way.  Page numbers, headings, and similar things are attached
after each page has been ejected, by a special sequence of \TeX\ commands
called the current {\sl^{output routine}}.

Plain \TeX\ has an output routine that takes care of ordinary jobs. It
handles the simple things that most manuscripts require, and it also
copes with more complicated things like the insertions made with
^|\footnote| and ^|\topinsert|, as described in the dangerous bends of
Chapter~15. We shall begin the present chapter by discussing how to make
simple changes to the behavior of plain \TeX's output routine; then we
shall turn to the details of how to define output routines that do more
complex tasks.

If you run \TeX\ without modifying the ^{plain \TeX\ format}, you get
^^{page format, modifying}
pages that are numbered at the bottom; and each page will be approximately
8$1\over2$~inches wide and 11~inches tall, including 1-inch margins at
all four sides. This format is suitable for preprints of
technical papers, but you might well want to change it, especially if
you are not using \TeX\ to make a preprint of a technical paper.

For example, we saw in the experiments of Chapter 6 that the width of the
material on a page can be changed by giving a different value to the
horizontal line size, ^|\hsize|. Plain \TeX\ format says `|\hsize=6.5in|',
in order to obtain 8.5-inch pages with 1-inch margins; you can change
|\hsize| to whatever you want. Similarly, you can control the vertical
size of a page by changing ^|\vsize|. Plain \TeX\ sets |\vsize=8.9in|
(not |9in|, since |\vsize| doesn't include the space for page numbers at the
bottom of each page); if you say `|\vsize=4in|' you will get shorter pages,
with only 4 inches of copy per sheet. It's best not to monkey with
|\hsize| and |\vsize| except at the very beginning of a job, or after you
have ejected all pages from \TeX's memory.

If you want your output to be positioned differently when it is ultimately
printed, you can offset it by giving nonzero values to ^|\hoffset| and
^|\voffset|. For example,
\begintt
\hoffset=.5in   \voffset=1.5in
\endtt
will move the output half an inch to the right of its normal position, and
1.5 inches down. You should be careful not to offset the output so much
that it falls off the edge of the physical medium on which it is being
printed, unless you know that such out-of-bounds activity won't cause trouble.

\TeX\ is often used to typeset announcements, ^{brochures}, or other
documents for which ^{page numbers} are inappropriate. If you say
\begintt
\nopagenumbers
\endtt
at the beginning of your manuscript, plain \TeX\ will refrain from inserting
numbers at the bottom of each page.

\danger In fact, ^|\nopagenumbers| is a special case of a much more general
mechanism by which you can control headings and footings. The plain \TeX\
output routine puts out a special line of text called the {\sl^{headline}\/}
at the top of each page, and another special line of text called the
{\sl^{footline}\/} at the bottom. The headline is normally blank, and
the footline is normally a centered page number, but you can specify any
headline and footline that you want by changing the token lists
^|\headline| and ^|\footline|. For example, ^^|\hrulefill|
\begintt
\headline={\hrulefill}
\endtt
will put a horizontal rule at the top of every page. The basic idea is
that plain \TeX\ puts `|\line{\the\headline}|' at the top and
`|\line{\the\footline}|' at the bottom, with blank lines separating these
extra lines from the other material. \ (Recall that ^|\line| is an
abbreviation for `|\hbox to\hsize|'; hence the headline and footline are
put into boxes as wide as the normal lines on the page itself.) \ The
normal value of\/ |\headline| is `|\hfil|', so that no heading is visible.
The |\nopagenumbers| macro described earlier is simply an abbreviation for
`|\footline={\hfil}|'.

\danger The normal value of\/ |\footline| is `|\hss\tenrm\folio\hss|';
this centers the page number on a line, using font ^|\tenrm|, because
^|\folio| is a control sequence that produces the number of the current
page in text form.

\danger The page number appears in \TeX's internal register |\count0|,
as explained in Chapter~15, and plain \TeX\ makes ^|\pageno| an
abbreviation for ^|\count0|. Thus you can say `|\pageno=100|' if you want
the next page of your output to be number~100. The |\folio| macro
converts negative page numbers to ^{roman numerals}; if your manuscript
begins with `|\pageno=-1|', the pages will be numbered i, ii, iii, iv,
v,~etc. In fact, Appendix~B defines |\folio| to be an abbreviation for
^^|\romannumeral|^^|\number|
\begintt
\ifnum\pageno<0 \romannumeral-\pageno \else\number\pageno \fi
\endtt

\danger It is important to include the name of each font explicitly
whenever you are defining a headline or footline, because an output routine
in \TeX\ can come into action at somewhat unpredictable times. For example,
suppose that |\footline| had been set to `|\hss\folio\hss|', without
specifying |\tenrm|; then the page number would be typeset in whatever
font happens to be current when \TeX\ decides to output a page.
Mysterious effects can occur in such cases, because \TeX\ is typically
in the midst of page~101 when it is outputting page~100.

\dangerexercise Explain how to put ^{en-dashes} around the page numbers in a
plain \TeX\ job. For example, `\hbox{ -- 4 -- }' should appear at the bottom
of page~4.
\answer |\footline={\hss\tenrm-- \folio\ --\hss}|

\danger Here is an example of a headline in which the page numbers
appear at the top. Furthermore, odd-numbered and even-numbered pages are
treated differently:
\begintt
\nopagenumbers % suppress footlines
\headline={\ifodd\pageno\rightheadline \else\leftheadline\fi}
\def\rightheadline{\tenrm\hfil RIGHT RUNNING HEAD\hfil\folio}
\def\leftheadline{\tenrm\folio\hfil LEFT RUNNING HEAD\hfil}
\voffset=2\baselineskip
\endtt
English-language books traditionally have ^{odd-numbered pages} on the
right and ^{even-numbered pages} on the left. Text that appears as a
headline on several pages is often called a ``^{running head}.'' When
you use headlines, it is generally wise to set ^|\voffset| to the
equivalent of two lines of text, as shown in this example, so that there
will still be a margin of one inch at the top of your output pages.

\dangerexercise Suppose that you're using \TeX\ to typeset your ^{r\'esum\'e},
which is several pages long. Explain how to define |\headline| so that
the first page is headed by `{\bf R\'ESUM\'E}', centered in boldface type,
while each subsequent page has a headline like this: ^^{Thor}
$$\line{R\'esum\'e of A. U. Thor \dotfill\ Page 2}$$
\answer |\headline={\ifnum\pageno=1 \hss\tenbf R\'ESUM\'E\hss|\parbreak
        |  \else\tenrm R\'esum\'e of A. U. Thor \dotfill\ Page \folio\fi}|
\smallskip\noindent (You should also say |\nopagenumbers| and
|\voffset=2\baselineskip|.)

\danger If you don't change the |\vsize|, all of the headlines and footlines
will occur in the same place regardless of the contents of the page
between them. Thus, for example, if you are using ^|\raggedbottom| as
explained in Chapter~15, so that pages do not always contain the same
amount of text, the raggedness will occur above the footline; the
footline won't move up. If you do change ^|\vsize|, the footline position
will change correspondingly, while the headline will stay put.

\ddanger The rest of this chapter is intended for people who want an
output format that is substantially different from what plain \TeX\ provides.
Double dangerous bends are used in all of the subsequent paragraphs, because
you should be familiar with the rest of \TeX\ before you plunge into these
final mysteries of the language. Chapter~22 taught you how to be a \TeX\
Master, i.e., a person who can produce complicated tables using |\halign|
and |\valign|; the following material will take you all the way to the
rank of ^{Grandmaster}, i.e., a person who can design output routines.
When you are ready for this rank, you will be pleased to discover
that---like alignments---output routines are not really so mysterious as
they may seem at first.

\ninepoint % it's all \ddangerous from here on
\ddanger Let's begin by recapping some of the rules at the end of Chapter~15.
\TeX\ periodically chooses to output a page of information, by breaking
its main vertical list at what it thinks is the best place, and at such times
it enters internal vertical mode and begins to read the commands in the
current |\output| routine. When the output routine begins, ^|\box255|
contains the page that \TeX\ has completed; the output routine is supposed to
do something with this vbox. When the output routine ends, the list of
items that it has constructed in internal vertical mode is placed just
before the material that follows the page break. In this way \TeX's
page-break decisions can effectively be changed: Some or all of the material
on the broken-off page can be removed and carried forward to the next page.

\ddanger The current ^|\output| routine is defined as a token list
parameter, just like ^|\everypar| or ^|\errhelp|, except that \TeX\
retains the begin-group symbol~`|{|' at the beginning
and the end-group symbol~`|}|' at the end. These ^{grouping characters}
help to keep the output routine from interfering with what
\TeX\ was doing when the page break was chosen; for example, an output
routine often changes the ^|\baselineskip| when it puts a headline
or footline on a page, and the extra ^{braces} keep this change local.
If no |\output| routine has been specified, or if the user has said
`|\output={}|', \TeX\ supplies its own routine, which is essentially
equivalent to `|\output={\shipout\box255}|'; this outputs the page without
any headline or footline, and without changing the page number.
^^{default output routine}^^|\shipout|

\ddanger \TeX's primitive command |\shipout|\<box> is what actually
causes output. It sends the contents of the box to the |dvi| file,
which is \TeX's main output file; after \TeX\ has finished,
the ^|dvi| file will contain a compact device-independent encoding of
instructions that specify exactly what should be printed.  When a
box is shipped out, \TeX\ displays the values of\/ |\count0| through
|\count9| on your terminal, ^^|\count0| as explained in Chapter~15; these
ten counters are also recorded in the |dvi| file, where they can be used
to identify the page. All of the ^|\openout|, ^|\closeout|, and ^|\write|
commands that appear inside of the \<box> are performed in their natural
order as that box is being shipped out. Since a |\write| command
expands macros, as explained in Chapter~21, \TeX's scanning mechanism
might detect syntax errors while a |\shipout| is in progress. If
^|\tracingoutput| is nonzero at the time of a |\shipout|, the contents
of the \<box> being shipped are written into your log file in symbolic
form. You can say |\shipout| anywhere, not only in an output routine.

\ddanger The delayed aspect of\/ |\write| imposes a noteworthy restriction:
It is necessary to be sure that all macros that might appear within the
text of a |\write| are properly defined when a |\shipout| command is
given. For example, the plain \TeX\ format in Appendix~B temporarily makes
spaces active and says `|\global\let|\]|=|^|\space|'; the reason is that
^|\obeyspaces| might be in force during a |\write| command, so a definition
for \] as an active character should exist during the next |\shipout|,
even though \TeX\ might no longer be making ^{spaces active} at that time.

\ddanger Chapter 15 points out that \TeX\ gives special values to
certain internal registers and parameters, in addition to |\box255|,
just before the output routine begins. Insertions are put into their
own vboxes, and ^|\insertpenalties| is set equal to the total number of
held-over insertions; furthermore the ^|\outputpenalty| parameter is
set to the value of the penalty at the current breakpoint. An output
routine can be made to do special things when these quantities have
special values. For example, the output routine of plain \TeX\ recognizes
a ^|\supereject| (which ejects all held-over insertions) by the fact that
|\supereject| causes |\outputpenalty| to be $-20000$, and by using
|\insertpenalties| to decide if any insertions are being held over.

\ddanger The default output routine, `|\shipout\box255|', illustrates one
extreme in which nothing is put into the vertical list that is carried over
to the next page. The other extreme is
\begintt
\output={\unvbox255
  \ifnum\outputpenalty<10000 \penalty\outputpenalty\fi}
\endtt
which ships nothing out and puts {\sl everything\/} back onto the main
vertical list. \ (The command `^|\unvbox||255|' takes the completed page
out of its box, and the command `|\penalty\outputpenalty|' reinserts the
penalty at the chosen breakpoint.) \ This makes a seamless join between
the completed page and the subsequent material, because \TeX\ has still
not discarded glue and penalties at the breakpoint when it invokes
an |\output| routine; hence \TeX\ will go back and reconsider the page
break. If the |\vsize| hasn't changed, and if all insertions have been
held in place, the same page break will be found; but it will be found
much faster than before, because the vertical list has already been
constructed---the paragraphing doesn't need to be done again. Of course,
an output routine like this makes \TeX\ spin its wheels endlessly, so
it~is of no use except as an example of an extreme case.

\ddanger To prevent such looping, your output routine should always
make progress of some sort whenever it comes into play. If you make a mistake,
\TeX\ may be able to help you diagnose the error, because a special
loop-detection mechanism has been built in:
There is an internal integer variable called ^|\deadcycles|,
which is cleared to zero after every |\shipout| and increased by~1
just before every |\output|. Thus, |\deadcycles| keeps track of how
many times an output routine has been initiated since the most recent
|\shipout|, unless you change the value of\/ |\deadcycles| yourself.
There's also an integer parameter called |\maxdeadcycles|, which plain
\TeX\ sets to~25. If\/ |\deadcycles| is greater than or equal to
|\maxdeadcycles| when your output routine is about to be started (i.e.,
when |\deadcycles| is about to be increased), \TeX\ issues an error
message and performs the ^{default output routine} instead of yours.

\ddanger When your output routine is finished, |\box255| should be void.
In other words, you must do something with the information in that
box; it should either be shipped out or put into some other place.
Similarly, ^|\box255| should be void when \TeX\ is getting ready to
fill it with a new page of material, just before starting an output
routine. If\/ |\box255| is nonvoid at either of those times, \TeX\ will
complain that you are misusing this special register, and the
register contents will be destroyed.

\ddanger But let's not talk forever about borderline cases and special
parameters; let's look at some real examples. The output routine of
plain \TeX, found in Appendix~B\null, is set up by saying
`|\output={\plainoutput}|', where ^|\plainoutput| is an abbreviation for
\begintt
\shipout\vbox{\makeheadline
  \pagebody
  \makefootline}
\advancepageno
\ifnum\outputpenalty>-20000 \else\dosupereject\fi
\endtt
Let us consider this ``program'' one line at a time:\enddanger

\medskip
\textindent{1)} The ^|\makeheadline| macro constructs a vbox of height
and depth zero in such a way that the headline is properly positioned
above the rest of the page. Its actual code is ^^|\headline|^^|\nointerlineskip|
\begintt
\vbox to 0pt{\vskip-22.5pt
  \line{\vbox to8.5pt{}\the\headline}\vss}
\nointerlineskip
\endtt
The magic constant $-22.5\pt$ is equal to $\bigl(\hbox{topskip}-\hbox{height of
strut}-2(\hbox{baselineskip})\bigr)$, i.e., $10\pt-8.5\pt-24\pt$;
^^{strut}^^|\vss| ^^|\line| this places the reference point of the headline
exactly $24\pt$ above the reference point of the top line on the page,
unless the headline or the top line are excessively large.

\medbreak
\textindent{2)} The ^|\pagebody| macro is an abbreviation for
\begintt
\vbox to\vsize{\boxmaxdepth=\maxdepth \pagecontents}
\endtt
The value of\/ ^|\boxmaxdepth| is set to ^|\maxdepth| so that the vbox
will be constructed under the assumptions that \TeX's page builder has
used to set up |\box255|.

\medbreak
\textindent{3)} The ^|\pagecontents| macro produces a vertical list for
everything that belongs on the main body of the page, namely the contents
of\/ |\box255| together with illustrations (inserted at the top)
and footnotes (inserted at the bottom): ^^|\topins| ^^|\footnote|
\begintt
\ifvoid\topins \else\unvbox\topins\fi
\dimen0=\dp255 \unvbox255
\ifvoid\footins\else % footnote info is present
  \vskip\skip\footins
  \footnoterule
  \unvbox\footins\fi
\ifraggedbottom \kern-\dimen0 \vfil \fi
\endtt
Here ^|\topins| and ^|\footins| are the insertion class numbers for
the two kinds of insertions used in plain \TeX; if more classes of
^{insertions} are added, |\pagecontents| should be changed accordingly.
Notice that the boxes are unboxed so that the glue coming from
insertions can help out the glue on the main page. The ^|\footnoterule|
macro in Appendix~B places a dividing line between the page and its
footnotes; it makes a net contribution of $0\pt$ to the height of
the vertical list. ^{Ragged-bottom setting} is achieved by inserting ^^|\vfil|
^{infinite glue}, which overpowers the stretchability of\/ ^|\topskip|.

\medbreak
\textindent{4)} The ^|\makefootline| macro puts ^|\footline| into
its proper position:
\begintt
\baselineskip=24pt \lineskiplimit=0pt
\line{\the\footline}
\endtt

\medbreak
\textindent{5)} The ^|\advancepageno| macro normally advances ^|\pageno|
by~$+1$; but if\/ |\pageno| is negative (for roman numerals), the advance is
by~$-1$. The new value of\/ |\pageno| will be appropriate for the next time
the output routine is called into action. ^^|\global|^^|\advance|
\begintt
\ifnum\pageno<0 \global\advance\pageno by-1
\else \global\advance\pageno by 1 \fi
\endtt

\medbreak
\textindent{6)} Finally, the ^|\dosupereject| macro is designed to clear
out any insertions that have been held over, whether they are illustrations
or footnotes or both: ^^|\insertpenalties| ^^|\supereject|
\begintt
\ifnum\insertpenalties>0
  \line{} \kern-\topskip \nobreak
  \vfill\supereject\fi
\endtt
The mysterious negative ^|\kern| here cancels out the natural space of the
^|\topskip| glue that goes above the empty |\line|; that empty line box
prevents the ^|\vfill| from disappearing into a page break.  The vertical
list that results from |\dosupereject| is placed on \TeX's list of things
to put out next, just after the straggling insertions have been
reconsidered as explained in Chapter~15. Hence another super-eject will
occur, and the process will continue until no insertions remain.

\ddangerexercise Explain how to change the output routine of plain \TeX\
so that it will produce twice as many pages. The material that would
ordinarily go on pages 1,~2, 3,~etc., should go onto pages 1,~3, 5,~\dots;
and the even-numbered pages should be entirely blank except for the
headline and footline. \ (Imagine that photographs will be mounted
on those blank pages later.)
\answer |\output={\plainoutput\blankpageoutput}|\parbreak
        |\def\blankpageoutput{\shipout\vbox{\makeheadline|\parbreak
        |    \vbox to\vsize{}\makefootline}\advancepageno}|

\ddanger Suppose now that ^{double-column} format is desired.
More precisely, let's attempt to modify plain \TeX\ so that it sets type
in columns whose width is ^|\hsize||=3.2in|. Each actual page of
output should contain two such columns separated by $0.1\rm\,in$ of
space; thus the text area of each page will still be 6.5~inches wide.
The headlines and footlines should span both columns, but the columns
themselves should contain independent insertions as if they were the
facing pages of a book. In other words, each column should contain
its own footnotes and its own illustrations; we do not have to change
the ^|\pagebody| macro. ^^{two-column format} ^^{multicolumn format}

\ddanger In order to solve this problem, let us first introduce a
new dimension register called ^|\fullhsize| that represents the
width of an entire page.
\begintt
\newdimen\fullhsize
\fullhsize=6.5in \hsize=3.2in
\def\fullline{\hbox to\fullhsize}
\endtt
The ^|\makeheadline| and ^|\makefootline| macros should be modified
so that they use `^|\fullline|' instead of `^|\line|'.

\ddanger The new output routine will make use of a control sequence
|\lr| that is set to either `|L|' or `|R|', according as the
next column belongs at the left or at the right of the next page.
When a left column has been completed, the output routine simply
saves it in a box register; when a right column has been completed,
the routine outputs both columns and increases the page number.
^^|\advancepageno|
\begintt
\let\lr=L \newbox\leftcolumn
\output={\if L\lr
    \global\setbox\leftcolumn=\columnbox \global\let\lr=R
  \else \doubleformat \global\let\lr=L\fi
  \ifnum\outputpenalty>-20000 \else\dosupereject\fi}
\def\doubleformat{\shipout\vbox{\makeheadline
    \fullline{\box\leftcolumn\hfil\columnbox}
    \makefootline}
  \advancepageno}
\def\columnbox{\leftline{\pagebody}}
\endtt
The |\columnbox| macro uses |\leftline| in order to ensure that it produces a
box whose width is |\hsize|. The width of\/ |\box255| is usually, but not
always, equal to |\hsize| at the beginning of an output routine; any other
width would louse up the format.

\ddanger When double-column setting ends, there's a 50-50 chance that the
final column has fallen at the left, so it will not yet have been output.
The code
\begintt
\supereject
\if R\lr \null\vfill\eject\fi
\endtt
supplies an empty right-hand column in this case, ensuring that all
of the accumulated material will be printed. It's possible to do
fancier column balancing on the last page, but the details are tricky
if footnotes and other insertions need to be accommodated as well.
Appendix~E includes the macros that were used to balance the columns at the
end of the index in Appendix~I\null, and to start two-column format in mid-page.

\ddangerexercise How should the example above be modified if you want
^{three-column output}?
\answer Set |\hsize=2.1in|, allocate `|\newbox\midcolumn|', and use the
following code:
\begintt
\output={\if L\lr
    \global\setbox\leftcolumn=\columnbox \global\let\lr=M
  \else\if M\lr
    \global\setbox\midcolumn=\columnbox \global\let\lr=R
  \else \tripleformat \global\let\lr=L\fi\fi
  \ifnum\outputpenalty>-20000 \else\dosupereject\fi}
\def\tripleformat{\shipout\vbox{\makeheadline
    \fullline{\box\leftcolumn\hfil\box\midcolumn\hfil\columnbox}
    \makefootline}
  \advancepageno}
\endtt
At the end, |\supereject| and say `|\if L\lr \else\null\vfill\eject\fi|'
twice.

\ddanger Since \TeX's output routine lags behind its page-construction
activity, you can get erroneous results if you change the |\headline| or the
|\footline| in an uncontrolled way. For example, suppose that you are
typesetting a book, and that the format you are using allows chapters
to start in the middle of a page; then it would be a mistake to change
the ^{running headline} at the moment you begin a new chapter, since the
next actual page of output might not yet include anything from the new
chapter. Consider also the task of typesetting a dictionary or a
membership roster; a well-designed reference book displays the current range
of entries at the top of each page or pair of pages, so that it is
easy for readers to thumb through the book when they are searching for
isolated words or names. But \TeX's asynchronous output mechanism
makes it difficult, if not impossible, to determine just what range
of entries is actually present on a page.

\ddanger Therefore \TeX\ provides a way to put ``^{marks}'' into a list;
these marks inform the output routine about the range of information on
each page. The general idea is that you can say ^^|\mark|
\begindisplay
|\mark{|\<mark text>|}|
\enddisplay
in the midst of the information you are typesetting, where the
\<mark text> is a token list that is expanded as in the commands
|\edef|, |\message|, etc. \TeX\ puts an internal representation of
the mark text into the list it is building; then later on, when a
completed page is packed into |\box255|, \TeX\ allows the output
routine to refer to the first and last mark texts on that page.

\ddanger The best way to think of this is probably to imagine that
\TeX\ generates an arbitrarily long vertical list of boxes, glue,
and other items such as penalties and marks. Somehow that long vertical
list gets divided up into pages, and the pages are made available to
the output routine, one at a time. Whenever a page is put in |\box255|,
\TeX\ sets up the value of three quantities that act essentially
like macros:\enddanger
\smallskip
\item\bull ^|\botmark| is the mark text most recently encountered on the
page that was just boxed;

\item\bull ^|\firstmark| is the mark text that was first encountered on
the page that was just boxed;

\item\bull ^|\topmark| has the value that |\botmark| had just before the
current page was boxed.
\smallbreak\noindent
Before the first page, all three of these are null, i.e., they expand
to nothing. When there is no mark on a page, all three are equal to
the previous |\botmark|.

\ddanger For example, suppose that your manuscript includes exactly four
marks, and that the pages are broken in such a way that |\mark{|$\alpha$|}|
happens to fall on page~2, |\mark{|$\beta$|}| and |\mark{|$\gamma$|}|
on page~4, and |\mark{|$\delta$|}| on page~5. Then
$$\halign{\indent\hfil#\hfil&&\qquad\hfil#\hfil\cr
On page&|\topmark| is&|\firstmark| is&|\botmark| is\cr
\noalign{\vskip2pt}
1&null&null&null\cr
2&null&$\alpha$&$\alpha$\cr
3&$\alpha$&$\alpha$&$\alpha$\cr
4&$\alpha$&$\beta$&$\gamma$\cr
5&$\gamma$&$\delta$&$\delta$\cr
6&$\delta$&$\delta$&$\delta$\cr}$$

\ddanger When you use a |\mark| command in vertical mode, \TeX\ puts a
mark into the main vertical list. When you use a |\mark| command in
horizontal mode, \TeX\ treats it as vertical mode material like
^|\vadjust| and ^|\insert|; i.e., after the paragraph has been broken
into lines, each mark will go into the main vertical list just after the
box for the line where that mark originally appeared. If
you use |\mark| in restricted horizontal mode, the mark may migrate out to
the enclosing vertical list in the same way that |\insert| and |\vadjust|
items do (see Chapter~24); but a mark that is locked too deeply inside
a box will not ^{migrate}, so it will never appear as a |\firstmark|
or |\botmark|.  Similarly, a |\mark| that occurs in internal vertical mode
goes into a vbox, and it is not accessible in the main vertical list.

\ddanger Chapter 15 discusses the ^|\vsplit| command, which allows you to
break up vertical lists by yourself. This operation sometimes provides
a useful alternative to \TeX's ordinary page-building mechanism.
For example, if you simply want to typeset some material in two columns
of equal height, you can put that material into a vbox, then |\vsplit|
the box into two pieces; no output routine is needed at all. The
|\vsplit| operation sets up the values of two macro-like quantities
that were not mentioned in Chapter~15: ^|\splitfirstmark| and
^|\splitbotmark| expand to the mark texts of the first and last
marks that appear in the vertical list that was split off by the most
recent |\vsplit| command. Both quantities are null if there were no
such marks. The values of\/ |\topmark|, |\firstmark|, |\botmark|,
|\splitfirstmark|, and |\splitbotmark| are global; i.e., they are
not affected by \TeX's ^{grouping} mechanism.

\ddanger Most dictionaries use the equivalent of\/ |\firstmark| and |\botmark|
to give ^{guide words} at the top of each pair of facing pages. For example,
if the definition of the word `type' starts on page~1387 and continues
onto page~1388, the guide word on page~1387 (a right-hand page)
will be `type'; but the guide word at the top of page~1388 (a left-hand page)
will be the next word in the dictionary (e.g., `typecast') even though
the top of page~1388 is about `type'.

\ddanger The dictionary scheme works fine for dictionaries, since a reader
should start reading each dictionary entry at its beginning. But a different
scheme is appropriate for a technical book like the author's {\sl^{Art
of Computer Programming}}, ^^{Knuth} where Section~1.2.8 (for example)
starts in the middle of page~78, but the top of page~78 contains
exercises 19--24 of Section~1.2.7. The headline at the top of page~78
refers to `1.2.7', because that will help somebody who is searching
for exercise 1.2.7--22. Notice that the dictionary convention would put
`1.2.8' at the top of page~78, but that would be appropriate only if
Section~1.2.8 had begun exactly at the top of that page.

\ddanger Continuing this example from {\sl The Art of Computer Programming},
let's suppose that the \TeX\ manuscript for Section~1.2.8 begins with
a macro call like
\begintt
\beginsection 1.2.8. Fibonacci Numbers.
\endtt
How should |\beginsection| be defined? Here is one attempt:
\begintt
\def\beginsection #1. #2.
  {\sectionbreak
  \leftline{\sectionfont #1. #2}
  \mark{#1}
  \nobreak\smallskip\noindent}
\endtt
The |\sectionbreak| macro should encourage \TeX\ either to break the page at the
current position, or to leave a goodly amount of blank space; e.g.,
|\sectionbreak| might be an abbreviation for `|\penalty-200| |\vskip18pt|
|plus4pt| |minus6pt|'. The |\beginsection| macro ends with commands that
suppress indentation of the first paragraph in the section. But the thing
that concerns us with respect to output routines is the |\mark| command
that follows |\leftline|.  In the example we have been considering, the
beginning of Section~1.2.8 would insert `|\mark{1.2.8}|' into the main
vertical list just after the box containing the title of that section.

\ddanger Is such a |\mark| adequate? Unfortunately, no, not even if we
assume for simplicity that at most one section begins on each page.
The page that contains the beginning of Section~1.2.8 will then have
|\topmark=1.2.7| and |\firstmark=1.2.8|, regardless of whether or not the
section starts at the very top of the page. What we want in this application
is a cross between |\topmark| and |\firstmark|: something that will reflect
the mark text that represents the state of affairs just after the first
line of the page. And \TeX\ doesn't provide that.

\ddanger The solution is to emit the |\mark| just before the |\sectionbreak|,
instead of just after the |\leftline|. Then |\topmark| will always reflect
the truth about the section that is current at the top line. \ (Think about it.)

\ddanger However, the format for {\sl The Art of Computer Programming\/}
is more complex than this. On left-hand pages, the section number in the
headline is supposed to reflect the situation at the top of the page,
as we have just discussed, but on right-hand pages it is supposed to refer
to the bottom of the page. Our solution to the previous problem made
|\topmark| correct for the top, but it can make |\botmark| incorrect
at the bottom. In order to satisfy both requirements, it is necessary
to pack more information into the marks. Here's one way to solve the problem:
\begintt
\def\beginsection #1. #2.
  {\mark{\currentsection \noexpand\else #1}
  \sectionbreak
  \leftline{\sectionfont #1. #2}
  \mark{#1\noexpand\else #1} \def\currentsection{#1}
  \nobreak\smallskip\noindent}
\def\currentsection{} % the current section number
\endtt
The idea is to introduce two marks, one just before the section break
and one just after the section has begun. Furthermore each mark has two
parts; the mark just before the potential break between Sections 1.2.7
and~1.2.8 is `|1.2.7\else 1.2.8|', while the one just after that potential
break is `|1.2.8\else 1.2.8|'. It follows that the section number corresponding
to the bottom of a page is the left component of\/ |\botmark|; the section
number corresponding to the top of a page is the right component of\/
|\topmark|.  The |\rightheadline| macro can make use of
`^|\iftrue||\botmark\fi|' to read the left component, and the
|\leftheadline| macro can say `^|\expandafter|^|\iffalse||\topmark\fi|' to
read the right component.

\ddangerexercise B. C. ^{Dull} used a construction very much like the one above,
but he put the second |\mark| just before the |\leftline| instead of
just after it. What went wrong?
\answer He forgot that ^{interline glue} is inserted automatically before
the |\leftline|; this permits a legal breakpoint between the |\mark| and
the |\leftline| box, according to the rules of page breaking in Chapter~15.
One cure would be to say ^|\nobreak| just after the |\mark|; but it's
usually best to put marks and ^{insertions} just {\sl after\/} boxes.

\ddangerexercise The marks in the previous construction have the form
`$\alpha$|\else|$\,\beta$', where $\alpha$ and $\beta$ are two independent
pieces of information. The `|\else|' makes it possible to select
either $\alpha$ or $\beta$ by means of\/ ^|\iftrue| and ^|\iffalse|.
Generalize this idea: Suppose that you have an application in which
marks are supposed to carry five independent pieces of information,
and that each mark has the form `$\alpha_0$|\or|$\,\alpha_1$|\or|%
$\,\alpha_2$|\or|$\,\alpha_3$|\or|$\,\alpha_4$'. Explain how to select
any one of the five $\alpha$'s from such a mark.
\answer Say, for example, |\ifcase2\expandafter\relax\botmark\fi| to
read part $\alpha_2$ of\/ |\botmark|. Another solution puts the five
components into five parameters of a macro, analogous to the method
used by |\inxcheck| later in this chapter; but the |\ifcase| approach
is usually more efficient, because it lets \TeX\ pass over the unselected
components at high speed.

\ddanger Let's conclude our discussion of output routines by considering an
application to indexes, such as the index to this manual that appears
^^{index marks} ^^{index example}
in Appendix~I\null. The most complicated entries in such an index will
look something like this:
\begindisplay
Main entry, 4, 6, 8--10, 12, 14--16,\cr
\qquad 18--22, 24--28, 30.\cr
\quad first subsidiary entry, 1--3, 6, 10--11,\cr
\qquad 15, 21, 24, 28.\cr
\quad second subsidiary entry, 1, 3, 6--7,\cr
\qquad  10, 15, 21, 25, 28, 31.\cr
\enddisplay
Main entries and subsidiary entries are typeset ^{ragged-right},
with two ems of hanging indentation after the first line; subsidiary entries
are indented one em on the first line. Our goal will be to typeset
such material from input that looks like this:
\begintt
\beginindex
...
Main entry, 4, 6, 8--10, 12, 14--16, 18--22, 24--28, 30.
\sub first subsidiary entry, 1--3, 6, 10--11, 15, 21, 24, 28.
\sub second subsidiary entry, 1, 3, 6--7, 10, 15, 21, 25, %
  28, 31.
...
\endindex
\endtt
where `|...|'\ stands for other entries. Each line of input normally specifies
one main entry or one subsidiary entry; if an entry is so long that it doesn't
fit on a single input line, `\]|%|' is typed at the end of the line so
that it merges with the following one.

\ddanger The interesting thing about this index problem is that it is
desirable to set up a system of marks so that the output routine can insert
special lines of text when an entry has been broken between columns or
pages. For example, if a page break occurs between any of the six lines
of typeset output shown above, the output routine should emit the special
line
\begindisplay
Main entry ({\it continued}\thinspace):
\enddisplay
and if a page break occurs within a subsidiary entry, an additional special
line
\begindisplay
\quad subsidiary entry ({\it continued}\thinspace):
\enddisplay
should also appear. The solution below produces marks so that |\botmark|
will be null if a break occurs between main entries; it will be
`|Main entry|' if a break occurs after lines 1, 2, or~4 of the six
example output lines; it will be `|Main entry\sub first subsidiary entry|'
if a break occurs after line~3 (within the first subsidiary entry); and it will
be `|Main entry\sub second subsidiary entry|' if a break occurs after line~5.

\ddanger The reader may wish to try solving this problem before looking at
the solution, because it will then be easier to appreciate the subtler
issues that are involved. \ (Go ahead: Try to define a macro |\beginindex|
that does the ragged-right setting and produces the specified marks. Turn
back to the previous page to study the problem carefully, before peeking
at the answer.)
^^|\everypar| ^^|\futurelet| ^^|\exhyphenpenalty| ^^|\raggedright|
^^|\hangindent|
\begintt
\def\beginindex{\begingroup
  \parindent=1em \maxdepth=\maxdimen
  \def\par{\endgraf \futurelet\next\inxentry}
  \obeylines \everypar={\hangindent 2\parindent}
  \exhyphenpenalty=10000 \raggedright}
\def\inxentry{\ifx\next\sub \let\next=\subentry
  \else\ifx\next\endindex \let\next=\vfill
  \else\let\next=\mainentry \fi\fi \next}
\def\endindex{\mark{}\break\endgroup}
\let\sub=\indent \newtoks\maintoks \newtoks\subtoks
\def\mainentry#1,{\mark{}\noindent
  \maintoks={#1}\mark{\the\maintoks}#1,}
\def\subentry\sub#1,{\mark{\the\maintoks}\indent
  \subtoks={#1}\mark{\the\maintoks\sub\the\subtoks}#1,}
\endtt
Even if you have read this solution, you probably want an explanation of
what it does, because it uses ``\TeX tics'' that have not appeared before
in this manual.\enddanger
\smallskip
\textindent{1)} The |\beginindex| macro uses ^|\begingroup| to keep other
changes local; thus, it won't be necessary to restore ^|\parindent| and
^|\maxdepth|, etc., to their former values when the index is finished.
The |\maxdepth| parameter is set to ^|\maxdimen|, which is essentially
infinite, so that |\box255| will have the true depth of the last box that
it contains; we will use this fact below. \ (It is safe to disable
|\maxdepth| in this way, since the entries in an index can be assumed
to have reasonably small depth.) \ Notice that ^|\obeylines| is used, so
that ^|\par| will effectively be inserted at the end of every line of
input. The meaning of\/ |\par| is changed so that it does more than usual:
First it does ^|\endgraf|, which is \TeX's ordinary |\par| operation;
then it sets |\next| to the first token of the next line, after which the
macro |\inxentry| will be expanded.

\smallskip
\textindent{2)} When |\inxentry| comes into play it looks at |\next|
to decide what to do. There are three cases: If\/ |\next| is `|\sub|',
the line will be treated as a subsidiary entry; if\/ |\next| is `|\endindex|',
the next commands executed will be `|\vfill|\allowbreak
|\mark{}|\allowbreak|\break|\allowbreak|\endgroup|';
otherwise the line will be treated as a main entry.

\smallskip
\textindent{3)} The text of a main entry is put into parameter |#1| of
|\mainentry|; this parameter is delimited by a comma. The first thing that
|\mainentry| does is `|\mark{}|', which clears the mark in case of a
break between entries. Then comes `^|\noindent|', which causes \TeX\ to
go into horizontal mode and to emit ^|\parskip| glue. \ (The |\parskip|
glue will be a legal breakpoint between lines; it will later be followed
by interline glue, when the first line of the main entry has been typeset
by \TeX's paragraphing routine.) \ Then another |\mark| is put into
the paragraph itself; this one contains the text of the main entry,
and a ^|\toks| register called ^|\maintoks| is used to ^{inhibit
expansion} of the mark text. When the paragraph is completed and broken
into lines, this particular mark will immediately follow the box for the
paragraph's first line, so it will be the |\botmark| if a page break
occurs anywhere within the paragraph.

\smallskip
\textindent{4)} A similar construction is used for |\subentry|, but the
mark is more complicated. The |\maintoks| register will still contain the
main entry. The text for the subsidiary entry is added using another
token list register, |\subtoks|.  Since |\sub| has been defined to equal
^|\indent|, it will not be expanded in this |\mark|.

\ddanger The macros just defined will typeset entries that contain the
necessary marks; now we must construct an output routine that uses
these marks in the desired way, to insert new lines that say `({\it
continued}\thinspace)' as mentioned above. Again, the reader is advised to
try solving this problem before looking at the following solution.
\begintt
\output={\dimen0=\dp255 \normaloutput
  \expandafter\inxcheck\botmark\sub\end}
\def\inxcheck#1\sub#2\end{\def\next{#1}%
  \ifx\next\empty % do nothing if \botmark is null
  \else\noindent #1\continued % `Main entry (continued):'
    \def\next{#2}%
    \ifx\next\empty % nothing more if \botmark has no \sub
    \else\let\sub=\continued \indent #2\fi
    \advance\dimen0 by-\prevdepth \kern\dimen0 \fi}
\def\continued{ ({\it continued}\thinspace):\endgraf}
\endtt
This coding is a bit more subtle than usual. It assumes that |\normaloutput|
takes care of shipping out |\box255| (possibly putting it into multicolumn
format) and advancing the page number; then comes new stuff, which is
performed by |\inxcheck|. The |\inxcheck| macro is invoked in an interesting
way that allows |\botmark| to be separated into its components. If\/ |\botmark|
^^{macro arguments}
is null, argument~|#1| to |\inxcheck| will be null; hence |\next| will be
found equivalent to ^|\empty|. \ (Plain \TeX\ says `|\def\empty{}|' in
order to accommodate situations like this.) \ If\/ |\botmark| doesn't contain
the token~|\sub|, argument~|#1| will be the contents of\/ |\botmark| while
|#2|~will be null. Otherwise, if\/ |\botmark| has the form $\alpha$|\sub|$\,
\beta$, argument~|#1| will be $\alpha$ and |#2|~will be `$\beta$|\sub|'.

\ddanger If\/ |\botmark| isn't null, the |\inxcheck| macro produces one or
more lines of text that will be contributed to \TeX's main vertical list
at the position of the page break. And here's where the most subtle point
arises: There will be ^{interline glue} at the page break, computed on
the basis of the depth of the box that preceded the break. That depth
is known to the output routine, since it's the depth of\/ |\box255|. \ (The
value of\/ ^|\maxdepth| was made infinite for precisely this reason.) \
Therefore the |\inxcheck| macro can insert a ^|\kern| to compensate for
the difference in depth between the old box and the one that will be
inserted before the interline glue that has already been computed. Without
this |\kern|, the spacing would be wrong. The reader should study this
example carefully, to understand the reasoning behind the |\kern| command,
before designing an output routine that inserts new boxes between
random lines of output.

\ddangerexercise Modify this construction so that continuation lines are
inserted only in the left columns of even-numbered pages, assuming
two-column format.
\answer |\output={\dimen0=\dp255 \normaloutput|\parbreak
        |  \ifodd\pageno\else\if L\lr|\parbreak
        |    \expandafter\inxcheck\botmark\sub\end\fi\fi}|
\smallskip\noindent In this case the |\normaloutput| macro should be the
two-column output routine that was described earlier in this chapter, beginning
with `|\if L\lr|' and ending with `|\let\lr=L\fi|'. \ (There is no need
to test for |\supereject|.)

\ddangerexercise True or false: The |\inxcheck| macro in this
example contributes at most two lines of output to the main vertical list.
\answer False. If the text of the main and/or subsidiary entry is lengthy,
a continuation line may actually become two or more lines. \ (Incidentally,
hanging indentation will then occur, because the |\everypar| command---which
was set up outside the |\output| routine---is effective inside.) \ The
|\vsize| must be large enough to accommodate all continuation lines plus
at least one more line of index material, or else infinite looping will occur.

\ddanger When \TeX\ sees an ^|\end| command, it terminates the job only
if the main vertical list has been entirely output and if\/ ^|\deadcycles||=0|.
Otherwise it inserts the equivalent of
\begintt
\line{} \vfill \penalty-'10000000000
\endtt
into the main vertical list, and prepares to read the `|\end|' token again.
This has the effect of invoking the output routine repeatedly until
everything has been shipped out. In particular, the last column of two-column
format will not be lost.

\ddanger It is possible to devise output routines that always leave a residue
on the main vertical list, yet they never allow |\deadcycles| to
increase. In such a case \TeX\ will never come to an end! An output
routine can recognize that it is being invoked by \TeX's endgame, because
of the highly negative |\outputpenalty| caused by the special
^|\penalty-'10000000000|. At such times the output routine should modify
its behavior, if necessary, so that a happy ending will ensue.

\endchapter

I think you will like them,
when you shall see them on a beautiful quarto page,
where a neat rivulet of text
shall meander through a meadow of margin.
'Fore Gad they will be the most elegant things of their kind!
\author RICHARD BRINSLEY ^{SHERIDAN}, %
  {\sl The School for Scandal\/} (1777) % Act I Sc 1

\bigskip

The influence of technical changes upon outputs
through variation in the general investment level {\cmman\char'14}
is so small that actually it could have been neglected.
\author WASSILY W. ^{LEONTIEF}\kern-1pt, %
 {\sl The Structure of American Economy, 1919--1929\/} (1941)

\eject
\beginchapter Chapter 24. Summary of\\Vertical\\Mode

^^{vertical mode}
The whole \TeX\ language has been presented in the previous chapters;
we have finally reached the end of our journey into previously
uncharted territory. Hurray! Victory! Now it is time to take a more
systematic look at what we have encountered: to consider the facts in an
orderly manner, rather than to mix them up with informal examples and
applications as we have been doing. A child learns to speak a language
before learning formal rules of grammar, but the rules of grammar come
in handy later on when the child reaches adulthood. The purpose of
this chapter---and of the two chapters that follow---is to present a
precise and concise summary of the language that \TeX\ understands, so
that mature users will be able to communicate as effectively as possible
with the machine.

We will be concerned in these chapters solely with \TeX's {\sl^{primitive}\/}
operations, rather than with the higher-level features of plain \TeX\ format
that most people deal with. Therefore novice users should put off reading
Chapters 24--26 until they feel a need to know what goes on inside the
computer. Appendix~B contains a summary of plain \TeX, together with a
ready-reference guide to the things that most people want to know about
\TeX\ usage. The best way to get an overview of \TeX\ from a high level
is to turn to the opening pages of Appendix~B.

\medskip\ninepoint
Our purpose here, however, is to survey the low-level parts of \TeX\ on
which higher-level superstructures have been built, in order to provide
a detailed reference for people who do need to know the details.
The remainder of this chapter is set in small type, like that of the
present paragraph, since it is analogous to material that is marked
``doubly dangerous'' in other chapters. Instead of using dangerous bend
signs repeatedly, let us simply agree that Chapters 24--26 are dangerous
by definition.

\medbreak
\TeX\ actually has a few features that didn't seem to be worth mentioning
in previous chapters, so they will be introduced here as part of our
complete survey. If there is any disagreement between something that was
said previously and something that will be said below, the facts in the
present chapter and its successors should be regarded as better
approximations to the ^{truth}.

\medbreak
We shall study \TeX's digestive processes, i.e., what \TeX\ does with
the lists of tokens that arrive in its ``stomach.'' Chapter~7 has
described the process by which input files are converted to lists of
tokens in \TeX's ``mouth,'' and Chapter~20 explained how expandable tokens
^^{anatomy of TeX}
are converted to unexpandable ones in \TeX's ``gullet'' by a process
similar to regurgitation. When unexpandable tokens finally reach \TeX's
gastro-intestinal tract, the real activity of typesetting begins, and
that is what we are going to survey in these summary chapters.

Each token that arrives in \TeX's tummy is considered to be a {\sl^{command}\/}
that the computer will obey.  For example, the letter `{\tt L}' is a command
to typeset an `L' in the current font; `|\par|' tells \TeX\ to
finish a paragraph.  \TeX\ is always in one of six modes, as
described in Chapter~13, and a command sometimes means different things
in different modes. The present chapter is about vertical mode (and
internal vertical mode, which is almost the same): We shall
discuss \TeX's response to every primitive command, when that command occurs
in vertical mode. Chapters 25 and~26 characterize horizontal mode and math mode
in a similar way, but those chapters are shorter than this one because
many commands have the same behavior in all modes; the rules for such
commands will not be repeated thrice, they will appear only once.

Some commands have ^{arguments}. In other words, one or more of the tokens
that follow a command might be used to modify that command's behavior,
and those tokens are not considered to be commands themselves. For example,
when \TeX\ processes the sequence of tokens that corresponds to
`|\dimen2=2.5pt|', it considers only the first token `|\dimen|' to be a
command; the next tokens are swept up as part of the operation, because \TeX\
needs to know what |\dimen| register is to be set equal to what \<dimen> value.

We shall define \TeX's parts of speech by using a modified form of the
grammatical notation that was introduced about 1960 by John ^{Backus} and
Peter ^{Naur} for the definition of computer languages. Quantities in
^{angle brackets} will either be explained in words or they will be defined
by {\sl^{syntax rules}\/} that show exactly how they are formed from other
quantities. For example,
\beginsyntax
<unit of measure>\is<optional spaces><internal unit>
  \alt<optional {\tt true}><physical unit>
\endsyntax
defines a \<unit of measure> to be either an occurrence of
\<optional spaces> followed by an \<internal unit>, or \<optional {\tt true}>
followed by \<physical unit>. The symbol `\is' in a syntax rule
means ``is defined to be,'' and `\alt' means ``or.''

Sometimes a syntax rule is ^{recursive}, in the sense that the right-hand
side of the definition involves the quantity being defined. For example,
the rule
\beginsyntax
<optional spaces>\is<empty>\alt<space token><optional spaces>
\endsyntax
defines the grammatical quantity called \<optional spaces> to be either
\<empty>, or a \<space token> followed by \<optional spaces>. The quantity
^\<empty> stands for ``nothing,'' i.e., for no tokens at all; hence the
syntax rule just given is a formalized way of saying that \<optional
spaces> stands for a sequence of zero or more spaces.

The alternatives on the right-hand side of a syntax rule need not consist
entirely of quantities in angle brackets. Explicit tokens can be used
as well. For example, the rule
\beginsyntax
<plus or minus>\is|+|$_{12}$\alt|-|$_{12}$
\endsyntax
says that \<plus or minus> stands for a ^{character token} that is either
a plus sign or a~minus sign, with category code~12.

We shall use a special convention for ^{keywords}, since the actual syntax
of a keyword is somewhat technical. Letters in typewriter type like
`\[pt]' will stand for
\begindisplay
\<optional spaces>\<p or P>\<t or T>,
\enddisplay
where \<p or P> denotes any non-active character token for either |p| or~|P|
(independent of the category code), and where \<t or T> is similar.

When a control sequence like `|\dimen|' is used in the syntax rules below,
it stands for any token whose current meaning is the same as the meaning
that |\dimen| had when \TeX\ started up. Other tokens can be given this
same meaning, using |\let| or |\futurelet|, and the meaning of the
control sequence |\dimen| itself may be redefined by the user, but
the syntax rules take no note of this; they just use `|\dimen|'
as a way of referring to a particular primitive command of \TeX. \ (This
notation is to be distinguished from `\cstok{dimen}', which stands
for the control sequence token whose actual name is |dimen|; see
Chapter~7.) ^^{boxed words}

Control sequences sometimes masquerade as characters, if their meaning has
been assigned by |\let| or |\futurelet|. For example, Appendix~B says
\begintt
\let\bgroup={   \let\egroup=}
\endtt
and these commands make ^|\bgroup| and ^|\egroup| act somewhat like left
and right ^{curly} ^{braces}.  Such control sequences are called ``^{implicit
characters}''; they are interpreted in the same way as characters, when
\TeX\ acts on them as commands, but not always when they appear in
arguments to commands.  For example, the command `|\let\plus=+|' does not
make |\plus| an acceptable substitute for the character token `|+|$_{12}$'
in the syntax rule for \<plus or minus> given above, nor does the command
`|\let\p=p|' make |\p| acceptable as part of the keyword \[pt]. When
\TeX's syntax allows both explicit and implicit characters, the rules
below will be careful to say so, explicitly.

The quantity ^\<space token>, which was used in the syntax of \<optional
spaces> above, stands for an explicit or implicit space. In other words,
it denotes either a character token of category~10, or a control sequence
or active character whose current meaning has been made equal to such a
token by |\let| or |\futurelet|.

It will be convenient to use the symbols `|{|', `|}|', and `|$|' to stand
for any explicit or implicit character tokens of the respective categories
1, 2, and~3, whether or not the actual character codes are braces or dollar
signs. Thus, for example, plain~\TeX's |\bgroup| is an example of a `|{|',
and so are the tokens `|{|$_1$' and `|(|$_1$'; but `|{|$_{12}$' is not.

The last few paragraphs can be summarized by saying that the alternatives
on the right-hand sides of \TeX's formal syntax rules are made from one
or more of the following things: (1)~syntactic quantities like
 \<optional spaces>; (2)~explicit character tokens like |+|$_{12}$;
(3)~keywords like \[pt]; (4)~control sequence names like |\dimen|;
or (5)~the special symbols |{|, |}|, |$|.

\medbreak
Let us begin our study of \TeX's syntax by discussing the precise meanings
of quantities like \<number>, \<dimen>, and \<glue> that occur frequently
as arguments to commands. The most important of these is \<number>,
which specifies an integer value. Here's exactly what a \<number> is:
\beginsyntax
<number>\is<optional signs><unsigned number>
<optional signs>\is<optional spaces>
  \alt<optional signs><plus or minus><optional spaces>
<unsigned number>\is<normal integer>\alt<coerced integer>
<normal integer>\is<internal integer>
  \alt<integer constant><one optional space>
  \alt|'|$_{12}$<octal constant><one optional space>
  \alt|"|$_{12}$<hexadecimal constant><one optional space>
  \alt|`|$_{12}$<character token><one optional space>
<integer constant>\is<digit>\alt<digit><integer constant>
<octal constant>\is<octal digit>\alt<octal digit><octal constant>
<hexadecimal constant>\is<hex digit>\alt<hex digit><hexadecimal constant>
<octal digit>\is|0|$_{12}$\alt|1|$_{12}$\alt|2|$_{12}$\alt|3|$_{12}$\alt%
    |4|$_{12}$\alt|5|$_{12}$\alt|6|$_{12}$\alt|7|$_{12}$
<digit>\is<octal digit>\alt|8|$_{12}$\alt|9|$_{12}$
<hex digit>\is<digit>\alt|A|$_{11}$\alt|B|$_{11}$\alt|C|$_{11}$\alt%
    |D|$_{11}$\alt|E|$_{11}$\alt|F|$_{11}$
  \alt|A|$_{12}$\alt|B|$_{12}$\alt|C|$_{12}$\alt%
    |D|$_{12}$\alt|E|$_{12}$\alt|F|$_{12}$
<one optional space>\is<space token>\alt<empty>
<coerced integer>\is<internal dimen>\alt<internal glue>
\endsyntax
The value of a \<number> is the value of the corresponding \<unsigned number>,
times~$-1$ for every minus sign in the \<optional signs>.
An ^{alphabetic constant} denotes the character code in a
^\<character token>; \TeX\ does not expand this token, which should either
be a (character~code, category~code) pair, or an active character, or
a control sequence whose name consists of a single character.
\ (See Chapter~20 for a complete list of all situations in which \TeX\ does
not expand tokens.) \ An \<integer constant> must not be immediately followed by
a \<digit>; in other words, if several digits appear consecutively, they
are all considered to be part of the same \<integer constant>. A similar
remark applies to the quantities \<octal constant> and \<hexadecimal constant>.
The quantity ^\<one optional space> is \<empty> only if it has to be;
i.e., \TeX\ looks for \<one optional space> by reading a token and backing
up if a \<space token> wasn't there.

\ddangerexercise Can you think of a reason why you might want `|A|$_{12}$'
to be a \<hex digit> even though the letter {\tt A} has category~11? \
(Don't worry if your answer is ``no.''\thinspace)
\answer If\/ |\cs| has been defined by ^|\chardef| or ^|\mathchardef|, \TeX\
uses ^{hexadecimal notation} when it expands ^|\meaning||\cs|, and it
assigns category~12 to each digit of the expansion. You might have an
application in which you want the last part of the expansion to be treated
as a \<number>. \ (This is admittedly an obscure reason.)

The definition of \<number> is now complete except for the three quantities
called \<internal integer>, \<internal dimen>, and \<internal glue>, which
will be explained later; they represent things like parameters and registers.
For example, |\count1| and |\tolerance| and |\hyphenchar\tenrm| are
internal integers; |\dimen10| and |\hsize| and |\fontdimen6\tenrm| are
internal dimensions; |\skip100| and |\baselineskip| and |\lastskip| are
internal glue values. An internal dimension can be ``coerced'' to be an
integer by assuming units of scaled points. For example, if\/ |\hsize=100pt|
^^{coerce <dimen> to <number>}
^^{coerce <glue> to <dimen>}
and if\/ |\hsize| is used in the context of a \<number>, it denotes the
integer value 6553600. Similarly, an internal glue value can be coerced to
be an integer by first coercing it to be a dimension (omitting the
stretchability and shrinkability), then coercing that dimension.

\smallskip
Let's turn now to the syntax for \<dimen>, and for \<mudimen> its cousin:
\beginsyntax
<dimen>\is<optional signs><unsigned dimen>
<unsigned dimen>\is<normal dimen>\alt<coerced dimen>
<coerced dimen>\is<internal glue>
<normal dimen>\is<internal dimen>\alt<factor><unit of measure>
<factor>\is<normal integer>\alt<decimal constant>
<decimal constant>\is|.|$_{12}$\alt|,|$_{12}$
  \alt<digit><decimal constant>
  \alt<decimal constant><digit>
<unit of measure>\is<optional spaces><internal unit>
  \alt<optional {\tt true}><physical unit><one optional space>
<internal unit>\is[em]<one optional space>\alt[ex]<one optional space>
  \alt<internal integer>\alt<internal dimen>\alt<internal glue>
<optional {\tt true}>\is[true]\alt<empty>
<physical unit>\is[pt]\alt[pc]\alt[in]\alt[bp]\alt[cm]\alt[mm]\alt[dd]\alt%
    [cc]\alt[sp]\vadjust{\vskip 3pt minus 2pt}
<mudimen>\is<optional signs><unsigned mudimen>
<unsigned mudimen>\is<normal mudimen>\alt<coerced mudimen>
<coerced mudimen>\is<internal muglue>
<normal mudimen>\is<factor><mu unit>
<mu unit>\is<optional spaces><internal muglue>\alt[mu]<one optional space>
\endsyntax
When `|true|' is present, the factor is multiplied by~1000 and divided by
the ^|\mag| parameter. Physical units are defined in Chapter~10; |mu| is
explained in Chapter~18.

\goodbreak
Encouraged by our success in mastering the precise syntax of the quantities
\<number>, \<dimen>, and \<mudimen>, let's tackle \<glue> and \<muglue>:
\beginsyntax
<glue>\is<optional signs><internal glue>
  \alt<dimen><stretch><shrink>
<stretch>\is[plus]<dimen>\alt[plus]<fil dimen>\alt<optional spaces>
<shrink>\is[minus]<dimen>\alt[minus]<fil dimen>\alt<optional spaces>
<fil dimen>\is<optional signs><factor><fil unit><optional spaces>
<fil unit>\is[fil]\alt<fil unit>[l]
<muglue>\is<optional signs><internal muglue>
  \alt<mudimen><mustretch><mushrink>
<mustretch>\is[plus]<mudimen>\alt[plus]<fil dimen>\alt<optional spaces>
<mushrink>\is[minus]<mudimen>\alt[minus]<fil dimen>\alt<optional spaces>
\endsyntax

\TeX\ makes a large number of internal quantities accessible so that a
format designer can influence \TeX's behavior. Here is a list of all
these quantities, except for the parameters (which will be listed later).
\beginsyntax
<internal integer>\is<integer parameter>\alt<special integer>\alt^|\lastpenalty|
  \alt<countdef token>\alt^|\count|<8-bit number>\alt<codename><8-bit number>
  \alt<chardef token>\alt<mathchardef token>\alt^|\parshape|\alt^|\inputlineno|
  \alt^|\hyphenchar|<font>\alt^|\skewchar|<font>\alt^|\badness|
<special integer>\is^|\spacefactor|\alt^|\prevgraf|
  \alt^|\deadcycles|\alt^|\insertpenalties|
<codename>\is^|\catcode|\alt^|\mathcode|
  \alt^|\lccode|\alt^|\uccode|\alt^|\sfcode|\alt^|\delcode|
<font>\is<fontdef token>\alt^|\font|\alt<family member>
<family member>\is<font range><4-bit number>
<font range>\is^|\textfont|\alt^|\scriptfont|\alt^|\scriptscriptfont|
<internal dimen>\is<dimen parameter>\alt<special dimen>\alt^|\lastkern|
  \alt<dimendef token>\alt^|\dimen|<8-bit number>
  \alt<box dimension><8-bit number>\alt^|\fontdimen|<number><font>
<special dimen>\is^|\prevdepth|\alt^|\pagegoal|\alt^|\pagetotal|
  \alt^|\pagestretch|\alt^|\pagefilstretch|\alt^|\pagefillstretch|
  \alt^|\pagefilllstretch|\alt^|\pageshrink|\alt^|\pagedepth|
<box dimension>\is^|\ht|\alt^|\wd|\alt^|\dp|
<internal glue>\is<glue parameter>\alt^|\lastskip|
  \alt<skipdef token>\alt^|\skip|<8-bit number>
<internal muglue>\is<muglue parameter>\alt^|\lastskip|
  \alt<muskipdef token>\alt^|\muskip|<8-bit number>
\endsyntax
A ^\<countdef token> is a control sequence token in which the control
sequence's current meaning has been defined by ^|\countdef|; the other
quantities ^\<dimendef token>, etc.,
^^\<skipdef token>^^\<muskipdef token>^^\<chardef token>^^\<mathchardef token>
^^\<toksdef token> are defined similarly. A \<fontdef
token> refers to a definition by ^|\font|, or it can be the predefined font
identifier called ^|\nullfont|.  When a \<countdef token> is used as an
internal integer, it denotes the value of the corresponding ^|\count|
register, and similar statements hold for \<dimendef token>, \<skipdef
token>, \<muskipdef token>. When a \<chardef token> or \<mathchardef
token> is used as an internal integer, it denotes the value in the
^|\chardef| or ^|\mathchardef| itself. An ^\<8-bit number> is a \<number>
whose value is between 0~and $2^8-1=255$; a ^\<4-bit number> is similar.
^^\<15-bit number> ^^\<27-bit number>

\TeX\ allows |\spacefactor| to be an internal integer only in horizontal
modes; |\prevdepth| can be an internal dimension only in vertical modes;
|\lastskip| can be \<internal muglue> only in math mode when the current
math list ends with a muglue item; and |\lastskip| cannot be \<internal
glue> in such a case. When |\parshape| is used as an internal integer, it
denotes only the number of controlled lines, not their sizes or
indentations.  The seven special dimensions |\pagetotal|, |\pagestretch|,
and so on are all zero when the current page contains no boxes, and
|\pagegoal| is |\maxdimen| at such times (see Chapter~15).

\smallskip
From the syntax rules just given, it's possible to deduce exactly what
happens to ^{spaces} when they are in the vicinity of numerical quantities:
\TeX\ allows a \<number> or \<dimen> to be preceded by arbitrarily many
spaces, and to be followed by at most one space; however, there is no
optional space after a \<number> or \<dimen> that ends with an unexpandable
control sequence. For example, if \TeX\ sees `|\space\space24\space\space|' when
it is looking for a \<number>, it gobbles up the first three spaces, but
the fourth one survives; similarly, one space remains when `|24pt\space\space|'
and `|\dimen24\space\space|' and `|\pagegoal\space|' are treated as
\<dimen> values.

\ddangerexercise Is `|24\space\space pt|' a legal \<dimen>?
\answer Yes; any number of spaces can precede any keyword.

\ddangerexercise Is there any difference between `|+\baselineskip|',
`|- -\baselineskip|', and `|1\baselineskip|', when \TeX\ reads
them as \<glue>?
\answer The first two have the same meaning; but the third coerces
|\baselineskip| to a \<dimen> by suppressing the stretchability
and shrinkability that might be present.

\ddangerexercise What \<glue> results from |"DD DDPLUS2,5 \spacefactor\space|,
assuming the conventions of plain \TeX, when |\spacefactor| equals 1000?
\answer The natural width is $221\rm\,dd$ (which \TeX\ rounds to
$15497423\rm\,sp$ and displays as |236.47191pt|).
The stretchability is $2500\rm\,sp$, since an
internal integer is coerced to a dimension when it appears as an
^^{coerce <number> to <dimen>}
\<internal unit>. The shrinkability is zero. Notice that the final |\space|
is swallowed up as part of the optional ^{spaces} of the \<shrink> part
in the syntax for \<glue>. \ (If |PLUS| had been |MINUS|, the final |\space|
would {\sl not\/} have been part of this \<glue>!)

Let's turn now to \TeX's ^{parameters}, which the previous chapters have
introduced one at a time; it will be convenient to assemble them
all together. An ^\<integer parameter> is one of the following tokens:
\begindisplay\belowdisplayskip=3pt plus 6pt \abovedisplayskip=3pt plus 1pt%
\openup.15pt
^|\pretolerance|\quad(badness tolerance before hyphenation)\cr
^|\tolerance|\quad(badness tolerance after hyphenation)\cr
^|\hbadness|\quad(badness above which bad hboxes will be shown)\cr
^|\vbadness|\quad(badness above which bad vboxes will be shown)\cr
^|\linepenalty|\quad(amount added to badness of every line in a paragraph)\cr
^|\hyphenpenalty|\quad(penalty for line break after discretionary hyphen)\cr
^|\exhyphenpenalty|\quad(penalty for line break after explicit hyphen)\cr
^|\binoppenalty|\quad(penalty for line break after binary operation)\cr
^|\relpenalty|\quad(penalty for line break after math relation)\cr
^|\clubpenalty|\quad(penalty for creating a club line at bottom of page)\cr
^|\widowpenalty|\quad(penalty for creating a widow line at top of page)\cr
^|\displaywidowpenalty|\quad(ditto, before a display)\cr
^|\brokenpenalty|\quad(penalty for page break after a hyphenated line)\cr
^|\predisplaypenalty|\quad(penalty for page break just before a display)\cr
^|\postdisplaypenalty|\quad(penalty for page break just after a display)\cr
^|\interlinepenalty|\quad(additional penalty for page break between lines)\cr
^|\floatingpenalty|\quad(penalty for insertions that are split)\cr
\noalign{\goodbreak}
^|\outputpenalty|\quad(penalty at the current page break)\cr
^|\doublehyphendemerits|\quad(demerits for consecutive broken lines)\cr
^|\finalhyphendemerits|\quad(demerits for a penultimate broken line)\cr
^|\adjdemerits|\quad(demerits for adjacent incompatible lines)\cr
^|\looseness|\quad(change to the number of lines in a paragraph)\cr
^|\pausing|\quad(positive if pausing after each line is read from a file)\cr
^|\holdinginserts|\quad(positive if insertions remain dormant in output box)\cr
^|\tracingonline|\quad(positive if showing diagnostic info on the terminal)\cr
^|\tracingmacros|\quad(positive if showing macros as they are expanded)\cr
^|\tracingstats|\quad(positive if showing statistics about memory usage)\cr
^|\tracingparagraphs|\quad(positive if showing line-break calculations)\cr
^|\tracingpages|\quad(positive if showing page-break calculations)\cr
^|\tracingoutput|\quad(positive if showing boxes that are shipped out)\cr
^|\tracinglostchars|\quad(positive if showing characters not in the font)\cr
^|\tracingcommands|\quad(positive if showing commands
   before they are executed)\cr
^|\tracingrestores|\quad(positive if showing deassignments when groups end)\cr
^|\language|\quad(the current set of hyphenation rules)\cr
^|\uchyph|\quad(positive if hyphenating words beginning with capital letters)\cr
^|\lefthyphenmin|\quad(smallest fragment at beginning of hyphenated word)\cr
^|\righthyphenmin|\quad(smallest fragment at end of hyphenated word)\cr
^|\globaldefs|\quad(nonzero if overriding |\global| specifications)\cr
^|\defaulthyphenchar|\quad(^|\hyphenchar| value when a font is loaded)\cr
^|\defaultskewchar|\quad(^|\skewchar| value when a font is loaded)\cr
^|\escapechar|\quad(escape character in the output of
   control sequence tokens)\cr
^|\endlinechar|\quad(character placed at the right end of an input line)\cr
^|\newlinechar|\quad(character that starts a new output line)\cr
^|\maxdeadcycles|\quad(upper bound on |\deadcycles|)\cr
^|\hangafter|\quad(hanging indentation changes after this many lines)\cr
^|\fam|\quad(the current family number)\cr
^|\mag|\quad(magnification ratio, times 1000)\cr
^|\delimiterfactor|\quad(ratio for variable delimiters, times 1000)\cr
^|\time|\quad(current time of day in minutes since midnight)\cr
^|\day|\quad(current day of the month)\cr
^|\month|\quad(current month of the year)\cr
^|\year|\quad(current year of our Lord)\cr
^|\showboxbreadth|\quad(maximum items per level when boxes are shown)\cr
^|\showboxdepth|\quad(maximum level when boxes are shown)\cr
^|\errorcontextlines|\quad(maximum extra context shown when errors occur)\cr
\enddisplay
The first few of these parameters have values in units of ``badness'' and
``penalties'' that affect line breaking and page breaking. Then come
demerit-oriented parameters; demerits are essentially given in units of
``badness squared,'' so those parameters tend to have larger values.
By contrast, the next few parameters (|\looseness|, |\pausing|, etc.)\
generally have quite small values (either $-1$ or 0 or 1 or~2).
Miscellaneous parameters complete the set. \TeX\ computes the date and time
when it begins a job, if the operating system provides such information; but
afterwards the clock does not keep ticking: The user can change |\time|
just like any ordinary parameter.  Chapter~10 points out that |\mag| must
not be changed after \TeX\ is committed to a particular magnification.

\goodbreak\noindent
A ^\<dimen parameter> is one of the following:
\begindisplay\openup.15pt
^|\hfuzz|\quad(maximum overrun before overfull hbox messages occur)\cr
^|\vfuzz|\quad(maximum overrun before overfull vbox messages occur)\cr
^|\overfullrule|\quad(width of rules appended to overfull boxes)\cr
^|\emergencystretch|\quad(reduces badnesses on final pass of line-breaking)\cr
^|\hsize|\quad(line width in horizontal mode)\cr
^|\vsize|\quad(page height in vertical mode)\cr
^|\maxdepth|\quad(maximum depth of boxes on main pages)\cr
^|\splitmaxdepth|\quad(maximum depth of boxes on split pages)\cr
^|\boxmaxdepth|\quad(maximum depth of boxes on explicit pages)\cr
^|\lineskiplimit|\quad(threshold where |\baselineskip| changes
   to |\lineskip|)\cr
^|\delimitershortfall|\quad(maximum space not covered by a delimiter)\cr
^|\nulldelimiterspace|\quad(width of a null delimiter)\cr
^|\scriptspace|\quad(extra space after subscript or superscript)\cr
^|\mathsurround|\quad(kerning before and after math in text)\cr
^|\predisplaysize|\quad(length of text preceding a display)\cr
^|\displaywidth|\quad(length of line for displayed equation)\cr
^|\displayindent|\quad(indentation of line for displayed equation)\cr
^|\parindent|\quad(width of\/ |\indent|)\cr
^|\hangindent|\quad(amount of hanging indentation)\cr
^|\hoffset|\quad(horizontal offset in |\shipout|)\cr
^|\voffset|\quad(vertical offset in |\shipout|)\cr
\enddisplay
And the possibilities for ^\<glue parameter> are:
\begindisplay\openup.15pt
^|\baselineskip|\quad(desired glue between baselines)\cr
^|\lineskip|\quad(interline glue if\/ |\baselineskip| isn't feasible)\cr
^|\parskip|\quad(extra glue just above paragraphs)\cr
^|\abovedisplayskip|\quad(extra glue just above displays)\cr
^|\abovedisplayshortskip|\quad(ditto, following short lines)\cr
^|\belowdisplayskip|\quad(extra glue just below displays)\cr
^|\belowdisplayshortskip|\quad(ditto, following short lines)\cr
^|\leftskip|\quad(glue at left of justified lines)\cr
^|\rightskip|\quad(glue at right of justified lines)\cr
^|\topskip|\quad(glue at top of main pages)\cr
^|\splittopskip|\quad(glue at top of split pages)\cr
^|\tabskip|\quad(glue between aligned entries)\cr
^|\spaceskip|\quad(glue between words, if nonzero)\cr
^|\xspaceskip|\quad(glue between sentences, if nonzero)\cr
^|\parfillskip|\quad(additional |\rightskip| at end of paragraphs)\cr
\enddisplay
Finally, there are three permissible ^\<muglue parameter> tokens:
\begindisplay\openup.15pt
^|\thinmuskip|\quad(thin space in math formulas)\cr
^|\medmuskip|\quad(medium space in math formulas)\cr
^|\thickmuskip|\quad(thick space in math formulas)\cr
\enddisplay
All of these quantities are explained in more detail somewhere else in
this book, and you can use Appendix~I to find out where.

\TeX\ also has parameters that are token lists. Such parameters do not
enter into the definitions of \<number> and such things, but we might as
well list them now so that our tabulation of parameters is complete.
A ^\<token parameter> is any of:
\begindisplay\openup.15pt
^|\output|\quad(the user's output routine)\cr
^|\everypar|\quad(tokens to insert when a paragraph begins)\cr
^|\everymath|\quad(tokens to insert when math in text begins)\cr
^|\everydisplay|\quad(tokens to insert when display math begins)\cr
^|\everyhbox|\quad(tokens to insert when an hbox begins)\cr
^|\everyvbox|\quad(tokens to insert when a vbox begins)\cr
^|\everyjob|\quad(tokens to insert when the job begins)\cr
^|\everycr|\quad(tokens
   to insert after every ^|\cr| or nonredundant ^|\crcr|)\cr
^|\errhelp|\quad(tokens that supplement an |\errmessage|)\cr
\enddisplay
That makes a total of 103 parameters of all five kinds.

\ddangerexercise Explain how |\everyjob| can be non-null when a job begins.
\answer If it was non-null when a ^|\dump| operation occurred. Here's
^^|\jobname|
a nontrivial example, which sets up ^|\batchmode| and puts ^|\end| at the
end of the input file:
\begintt
\everyjob={\batchmode\input\jobname\end}
\endtt

It's time now to return to our original goal, namely to study the commands
that are obeyed by \TeX's digestive organs. Many commands are
carried out in the same way regardless of the current mode. The most
important commands of this type are called {\sl^{assignments}}, since
they assign new values to the meaning of control sequences or to \TeX's
internal quantities. For example, `|\def\a{a}|' and `|\parshape=1 5pt 100pt|'
and `|\advance\count20 by-1|' and `|\font\ff = cmff at 20pt|' are all
assignments, and they all have the same effect in all modes.
Assignment commands often include an~^|=|~sign, ^^{equals sign}
but in all cases this sign is optional; you can leave it out if you
don't mind the fact that the resulting \TeX\ code might not look quite
like an assignment.
\beginsyntax
<assignment>\is<non-macro assignment>\alt<macro assignment>
<non-macro assignment>\is<simple assignment>
  \alt^|\global|<non-macro assignment>
<macro assignment>\is<definition>\alt<prefix><macro assignment>
<prefix>\is|\global|\alt^|\long|\alt^|\outer|
<equals>\is<optional spaces>\alt<optional spaces>\thinspace|=|$_{12}$
\endsyntax
This syntax shows that every assignment can be prefixed by |\global|, but
only macro-definition assignments are allowed to be prefixed by |\long|
or |\outer|. Incidentally, if the |\globaldefs| parameter is positive at the
time of the assignment, a prefix of\/ |\global| is automatically implied;
but if\/ ^|\globaldefs| is negative at the time of the assignment, a prefix
of\/ |\global| is ignored. If\/ |\globaldefs| is zero (which it usually~is),
the appearance or nonappearance of\/ |\global| determines whether or not
a global assignment is made.
\beginsyntax
<definition>\is<def><control sequence><definition text>
<def>\is^|\def|\alt^|\gdef|\alt^|\edef|\alt^|\xdef|
<definition text>\is<parameter text><left brace><balanced text><right brace>
\endsyntax
Here ^\<control sequence> denotes a token that is either a control sequence
or an active character; ^\<left brace> and ^\<right brace> are explicit
character tokens whose category codes are respectively of types 1 and~2.
The ^\<parameter text> contains no \<left brace> or \<right brace> tokens,
and it obeys the rules of Chapter~20. All occurrences of \<left brace>
and \<right brace> tokens within the ^\<balanced text> must be properly nested
like \hbox{parentheses}. A |\gdef| command is equivalent to |\global\def|,
and |\xdef| is equivalent to |\global\edef|. \TeX\ reads the
\<control sequence> and \<parameter text> tokens and the opening \<left brace>
without expanding them; it expands the \<balanced text>\allowbreak\<right
brace> tokens only in the case of\/ |\edef| and |\xdef|.

Several commands that we will study below have a syntax somewhat like
that of a definition, but the \<parameter text> is replaced by an
arbitrary sequence of spaces and `|\relax|' commands, and the
\<left brace> token can be implicit:
\beginsyntax
<filler>\is<optional spaces>\alt<filler>^|\relax|<optional spaces>
<general text>\is<filler>|{|<balanced text><right brace>
\endsyntax
The main purpose of a \<general text> is to specify the \<balanced text>
inside.

Many different kinds of assignments are possible, but they fall into
comparatively few patterns, as indicated by the following syntax rules:
\beginsyntax
<simple assignment>\is<variable assignment>\alt<arithmetic>
  \alt<code assignment>\alt<let assignment>\alt<shorthand definition>
  \alt<fontdef token>\alt<family assignment>\alt<shape assignment>
  \alt^|\read|<number>[to]<optional spaces><control sequence>
  \alt^|\setbox|<8-bit number><equals><filler><box>
  \alt^|\font|<control sequence><equals><filename><at clause>
  \alt<global assignment>
<variable assignment>\is<integer variable><equals><number>
  \alt<dimen variable><equals><dimen>
  \alt<glue variable><equals><glue>
  \alt<muglue variable><equals><muglue>
  \alt<token variable><equals><general text>
  \alt<token variable><equals><filler><token variable>
<arithmetic>\is^|\advance|<integer variable><optional {\tt by}><number>
  \alt|\advance|<dimen variable><optional {\tt by}><dimen>
  \alt|\advance|<glue variable><optional {\tt by}><glue>
  \alt|\advance|<muglue variable><optional {\tt by}><muglue>
  \alt^|\multiply|<numeric variable><optional {\tt by}><number>
  \alt^|\divide|<numeric variable><optional {\tt by}><number>
<optional {\tt by}>\is[by]\alt\<optional spaces>
<integer variable>\is<integer parameter>\alt<countdef token>
  \alt^|\count|<8-bit number>
<dimen variable>\is<dimen parameter>\alt<dimendef token>
  \alt^|\dimen|<8-bit number>
<glue variable>\is<glue parameter>\alt<skipdef token>
  \alt^|\skip|<8-bit number>
<muglue variable>\is<muglue parameter>\alt<muskipdef token>
  \alt^|\muskip|<8-bit number>
<token variable>\is<token parameter>\alt<toksdef token>
  \alt^|\toks|<8-bit number>
<numeric variable>\is<integer variable>\alt<dimen variable>
  \alt<glue variable>\alt<muglue variable>% I want to force a page break here!
\endgraf\penalty-500\syntaxrule% this defeats the \beginsyntax trickery
<at clause>\is[at]<dimen>\alt[scaled]<number>\alt<optional spaces>
<code assignment>\is<codename><8-bit number><equals><number>
<let assignment>\is^|\futurelet|<control sequence><token><token>
  \alt^|\let|<control sequence><equals><one optional space><token>
<shorthand definition>\is^|\chardef|<control sequence><equals><8-bit number>
  \alt^|\mathchardef|<control sequence><equals><15-bit number>
  \alt<registerdef><control sequence><equals><8-bit number>
<registerdef>\is^|\countdef|\alt^|\dimendef|\alt^|\skipdef|\alt%
    ^|\muskipdef|\alt^|\toksdef|
<family assignment>\is<family member><equals><font>
<shape assignment>\is^|\parshape|<equals><number><shape dimensions>
\endsyntax
The \<number> at the end of a \<code assignment> must not be negative, except in
the case that a |\delcode| is being assigned. Furthermore, that \<number> should
be at most 15~for |\catcode|, 32768~for |\mathcode|, 255~for |\lccode| or
|\uccode|, 32767~for |\sfcode|, and $2^{24}-1$~for |\delcode|. In a
\<shape assignment> for which the \<number> is $n$, the ^\<shape dimensions>
are \<empty> if $n\le0$, otherwise they consist of $2n$ consecutive
occurrences of \<dimen>. \TeX\ does not expand tokens when it scans the
arguments of\/ |\let| and |\futurelet|.

\ddangerexercise We discussed the distinction between explicit and ^{implicit
character tokens} earlier in this chapter. Explain how you can make the
control sequence |\cs| into an implicit space, using (a)~|\futurelet|,
(b)~|\let|.
\answer (a) |\def\\#1\\{}\futurelet\cs\\|\]|\\|. (b) |\def\\{\let\cs= }\\|\].
\ (There are many other solutions.)

All of the assignments mentioned so far will obey \TeX's grouping structure;
i.e., the changed quantities will be restored to their former values
when the current group ends, unless the change was global. The remaining
assignments are different, since they affect \TeX's global font tables
or hyphenation tables, or they affect certain control variables of such
^^{global parameters}
an intimate nature that grouping would be inappropriate. In all of the
following cases, the presence or absence of\/ |\global| as a prefix has no
effect.
\beginsyntax
<global assignment>\is<font assignment>
  \alt<hyphenation assignment>
  \alt<box size assignment>
  \alt<interaction mode assignment>
  \alt<intimate assignment>
<font assignment>\is^|\fontdimen|<number><font><equals><dimen>
  \alt^|\hyphenchar|<font><equals><number>
  \alt^|\skewchar|<font><equals><number>
<hyphenation assignment>\is^|\hyphenation|<filler>|{|<hyphenations>|}|
  \alt^|\patterns|<filler>|{|<patterns>|}|
<box size assignment>\is<box dimension><8-bit number><equals><dimen>
<interaction mode assignment>\is^|\errorstopmode|\alt^|\scrollmode|
  \alt^|\nonstopmode|\alt^|\batchmode|
<intimate assignment>\is<special integer><equals><number>
  \alt<special dimen><equals><dimen>
\endsyntax
When a |\fontdimen| value is assigned, the \<number> must be positive and
not greater than the number of parameters in the font's metric information
file, unless that font information has just been loaded into \TeX's
memory; in the latter case, you are allowed to increase the number of
parameters (see Appendix~F\null).  The \<special integer> and \<special
dimen> quantities were listed above when we discussed internal integers
and dimensions. When |\prevgraf| is set to a \<number>, the number must
not be negative.

The syntax for ^\<filename> is not standard in \TeX, because different
operating systems have different conventions. You should ask your local
system wizards for details on just how they have decided to implement file
names. However, the following principles should hold universally:
A~\<filename> should consist of \<optional spaces> followed by explicit
character tokens (after expansion). A sequence of six or fewer ordinary
letters and/or digits followed by a space should be a file name that works
in essentially the same way on all installations of\/ \TeX\null.  Uppercase
letters are not considered equivalent to their lowercase counterparts in
file names; for example, if you refer to fonts |cmr10| and |CMR10|, \TeX\
will not notice any similarity between them, although it might input the
same font metric file for both fonts.

\TeX\ takes precautions so that constructions like `|\chardef\cs=10\cs|' and
`|\font\cs=name\cs|' won't expand the second |\cs| until
the assignments are done.

Our discussion of assignments is complete except that the |\setbox|
assignment involves a quantity called \<box> that has not yet been
defined. Here is its syntax:
\beginsyntax
<box>\is^|\box|<8-bit number>\alt^|\copy|<8-bit number>
  \alt^|\lastbox|\alt^|\vsplit|<8-bit number>[to]<dimen>
  \alt^|\hbox|<box specification>|{|<horizontal mode material>|}|
  \alt^|\vbox|<box specification>|{|<vertical mode material>|}|
  \alt^|\vtop|<box specification>|{|<vertical mode material>|}|
<box specification>\is[to]<dimen><filler>
  \alt[spread]<dimen><filler>\alt<filler>
\endsyntax
The |\lastbox| operation is not permitted in math modes, nor is it allowed
in vertical mode when the main vertical list has been entirely contributed
to the current page. But it is allowed in horizontal modes and in
internal vertical mode; in such modes it refers to (and removes) the
last item of the current list, provided that the last item is an hbox or~vbox.

The three last alternatives for a \<box> present us with a new situation:
The ^\<horizontal mode material> in an |\hbox| and the
^\<vertical mode material> in a |\vbox| can't simply be swallowed up
in one command like an \<8-bit number> or a \<dimen>; thousands of
commands may have to be executed before that box is constructed and
before the |\setbox| command can be completed.

Here's what really happens: A command like
\begindisplay
|\setbox|\<number>|=\hbox to|\<dimen>|{|\<horizontal mode material>|}|
\enddisplay
causes \TeX\ to evaluate the \<number> and the \<dimen>, and to put those
values on a ``stack'' for safe keeping. Then \TeX\ reads the `|{|' (which
stands for an explicit or implicit begin-group character, as explained
earlier), and this initiates a new level of grouping. At this point
\TeX\ enters restricted horizontal mode and proceeds to execute commands
in that mode. An arbitrarily complex box can now be constructed;
the fact that this box is eventually destined for a |\setbox| command
has no effect on \TeX's behavior while the box is being built. Eventually,
when the matching `|}|' appears, \TeX\ restores values that were
changed by assignments in the group just ended; then it packages the hbox
(using the size that was saved on the stack), and completes the
|\setbox| command, returning to the mode it was in at the time of
the |\setbox|.

\smallbreak
Let us now consider other commands that, like assignments, are
obeyed in basically the same way regardless of \TeX's current mode.

\def\\{\smallbreak\textindent{$\bull$}}
\\^|\relax|.\enskip
This is an easy one: \TeX\ does nothing.

\\|}|.\enskip
This one is harder, because it depends on the current group. \TeX\ should
now be working on a group that began with |{|; and it knows why it
started that group. So it does the appropriate finishing actions, undoes
the effects of non-global assignments, and leaves the group.
At this point \TeX\ might leave its current mode and return to a mode that
was previously in effect.

\\^|\begingroup|.\enskip
When \TeX\ sees this command, it enters a group that must be terminated
by |\endgroup|, not by |}|. The mode doesn't change.

\\^|\endgroup|.\enskip
\TeX\ should currently be processing a group that began with |\begingroup|.
Quantities that were changed by non-global assignments in that group
are restored to their former values. \TeX\ leaves the group,
but stays in the same mode.

\\^|\show|\stretch\<token>,
\stretch\stretch\stretch^|\showbox|\stretch\<8-bit number>,
\stretch\stretch\stretch^|\showlists|,
\stretch\stretch\stretch^|\showthe|$\langle$internal quantity$\rangle$.\enskip
These commands are intended to help you figure out what \TeX\ thinks it
is doing. The tokens following |\showthe| should be anything that can
follow |\the|, as explained in Chapter~20.

\ddangerexercise Review the rules for what can follow |\the| in
Chapter~20, and construct a formal syntax that defines ^\<internal
quantity> in a way that fits with the other syntax rules we have been
discussing.
\answer \<internal quantity>\is\<internal integer>\alt
\<internal dimen>\parbreak\qquad\alt\<internal glue>\alt\<internal muglue>\alt
\<internal nonnumeric>\parbreak \<internal nonnumeric>\is\<token
variable>\alt \<font>

\\^|\shipout|\<box>.\enskip
After the \<box> is formed---possibly by constructing it explicitly and
changing modes during the construction, as explained for |\hbox| earlier---its
contents are sent to the ^|dvi|~file (see Chapter~23).

\\^|\ignorespaces|\stretch$\langle$optional spaces$\rangle$.
\TeX\ reads (and expands) tokens, doing nothing until reaching one that is
not a \<space token>.

\\^|\afterassignment|\<token>.\enskip
The \<token> is saved in a special place; it will be inserted
back into the input just after the next assignment command has been
performed. An assignment need not follow immediately; if another
|\afterassignment| is performed before the next assignment, the second one
overrides the first. If the next assignment is a ^|\setbox|, and if the
assigned \<box> is |\hbox| or |\vbox| or |\vtop|, the \<token> will be
inserted just after the |{| in the box construction, not after the |}|;
it will also come just before any tokens inserted by ^|\everyhbox| or
^|\everyvbox|.

\\^|\aftergroup|\<token>.\enskip
The \<token> is saved on \TeX's stack; it will be inserted back into the
input just after the current group has been completed and its local
assignments have been undone. If several |\aftergroup| commands occur
in the same group, the corresponding commands will be scanned in the
same order; for example, `|{\aftergroup\a\aftergroup\b}|' yields `|\a\b|'.

\\^|\uppercase|\<general text>, ^|\lowercase|\<general text>.\enskip
The \<balanced text> in the general text is converted to uppercase form
or to lowercase form using the |\uccode| or |\lccode| table,
as explained in Chapter~7; no expansion is done. Then \TeX\ will
read that \<balanced text> again.

\\^|\message|\<general text>, ^|\errmessage|\<general text>.\enskip
The balanced text (with expansion) is written on the user's terminal,
using the format of error messages in the case of\/ |\errmessage|.
In the latter case the ^|\errhelp| tokens will be shown if they are nonempty
and if the user asks for help.

\\^|\openin|\stretch$\langle$4-bit number$\rangle$\stretch\<equals>\stretch
\<filename>, \ ^|\closein|\stretch$\langle$4-bit number$\rangle$.
These commands open or close the specified input stream, for use in
|\read| assignments as explained in Chapter~20.

\\|\immediate\openout|\<4-bit number>\<equals>\<filename>\kern-.4pt,
|\immediate\closeout|\allowbreak\<4-bit number>.\enskip ^^|\immediate|
The specified output stream is opened or closed, for use in |\write|
commands, as explained in Chapter~21.

\\|\immediate\write|\<number>\<general text>.\enskip
The balanced text is written on the file that corresponds to the
specified stream number, provided that such a file is open. Otherwise it
is written on the user's terminal and on the log file. \ (See Chapter~21;
the terminal is omitted if the \<number> is negative.)

\medbreak
That completes the list of mode-independent commands, i.e., the commands
that do not directly affect the lists that \TeX\ is building.  When \TeX\
is in vertical mode or internal vertical mode, it is constructing a
vertical list; when \TeX\ is in horizontal mode or restricted horizontal
mode, it is constructing a horizontal list; when \TeX\ is in math mode
or display math mode, it is constructing---guess what---a math list. In
each of these cases we can speak of the ``current list''; and there are
some commands that operate in essentially the same way, regardless of the
mode, except that they deal with different sorts of lists:

\\^|\openout|\<4-bit number>\<equals>\<filename>,
^|\closeout|\<4-bit number>,
^|\write|\allowbreak\<number>\<general text>.\enskip
These commands are recorded into a ``whatsit'' item, which is appended to
the current list. The command will be performed later, during any
|\shipout| that applies to this list, unless the list is part of
a box inside ^{leaders}.

\\^|\special|\<general text>.\enskip
The balanced text is expanded and put into a ``whatsit'' item, which is
appended to the current list. The text will eventually appear in the
^|dvi|~file as an instruction to subsequent software (see Chapter~21).

\\^|\penalty|\<number>.\enskip
A penalty item carrying the specified number is appended to the current list.
In vertical mode, \TeX\ also exercises the page builder (see~below).

\\^|\kern|\<dimen>, ^|\mkern|\<mudimen>.\enskip
A kern item carrying the specified dimension is appended to the current list.
In vertical modes this denotes a vertical space; otherwise it denotes a
horizontal space. An |\mkern| is allowed only in math modes.

\\^|\unpenalty|, ^|\unkern|, ^|\unskip|.\enskip
If the last item on the current list is respectively of type penalty,
kern, or glue (possibly including ^{leaders}), that item is removed from
the list. However, like |\lastbox|, these commands are not permitted in
vertical mode if the main vertical list-so-far has been entirely contributed
to the current page, since \TeX\ never removes items from the current page.

\\^|\mark|\<general text>.\enskip
The balanced text is expanded and put into a mark item, which is appended to
the current list. The text may eventually become the replacement text
for ^|\topmark|, ^|\firstmark|, ^|\botmark|, ^|\splitfirstmark|, and/or
^|\splitbotmark|, if this mark item ever gets into a vertical list. \ (Mark
items can appear in horizontal lists and math lists, but they have no
effect until they ``migrate'' out of their list. The ^{migration process}
is discussed below and in Chapter~25.)

\\^|\insert|\<8-bit number>\<filler>|{|\<vertical mode material>|}|;
the \<8-bit number> must not be~255. \ The `|{|' causes \TeX\ to enter
internal vertical mode and a new level of grouping. When the matching~`|}|'
is sensed, the vertical list is put into an insertion item that is
appended to the current list using the values of\/ ^|\splittopskip|,
^|\splitmaxdepth|, and ^|\floatingpenalty| that were current in the group
just ended. \ (See Chapter~15.) \ This insertion item leads ultimately to
a page insertion only if it appears in \TeX's main vertical list, so it
will have to ``^{migrate}'' there if it starts out in a horizontal list or
a math list.  \TeX\ also exercises the page builder (see below), after an
|\insert| has been appended in vertical~mode.

\\^|\vadjust|\<filler>|{|\<vertical mode material>|}|.\enskip
This is similar to |\insert|; the constructed vertical list goes into an
adjustment item that is appended to the current list. However,
|\vadjust| is not allowed in vertical modes. When an adjustment item
migrates from a horizontal list to a vertical list, the vertical list
inside the adjustment item is ``unwrapped'' and put directly into
the enclosing list.

\medbreak
\centerline{$*\qquad*\qquad*$}
\medskip\noindent
Almost everything we have discussed so far in this chapter could equally
well have appeared in a chapter entitled ``Summary of Horizontal Mode''
or a chapter entitled ``Summary of Math Mode,'' because \TeX\ treats
all of the commands considered so far in essentially the same way
regardless of the current mode. Chapters 25 and~26 are going to be
a lot shorter than the present one, since it will be unnecessary
to repeat all of the mode-independent rules.

But now we come to commands that are mode-dependent;
we shall conclude this chapter by discussing what \TeX\ does with the
remaining commands, when in vertical mode or internal vertical mode.

One of the things characteristic of vertical mode is the page-building
operation described in Chapter~15. \TeX\ periodically takes material
that has been put on the main vertical list and moves it from the
``contribution list'' to the ``current page.'' At such times the output
routine might be invoked. We shall say that \TeX\ {\sl exercises
the ^{page builder}\/} whenever it tries to empty the current
contribution list. The concept of contribution list exists only in
the outermost vertical mode, so nothing happens when \TeX\ exercises
the page builder in internal vertical mode.

Another thing characteristic of vertical modes is the ^{interline glue}
that is inserted before boxes, based on the values of\/ |\prevdepth|
and ^|\baselineskip| and ^|\lineskip| and ^|\lineskiplimit| as explained
in Chapter~12. If a command changes ^|\prevdepth|, that fact is
specifically mentioned below. The |\prevdepth| is initially set to
$-1000\pt$, a special value that inhibits interline glue, whenever \TeX\
begins to form a vertical list, except in the case of\/ |\halign| and
|\noalign| when the interline glue conventions of the outer list continue
inside the inner one.

\\^|\vskip|\<glue>, ^|\vfil|, ^|\vfill|, ^|\vss|, ^|\vfilneg|.\enskip
A glue item is appended to the current vertical list.

\\\<leaders>\<box or rule>\<vertical skip>.\enskip
Here ^\<vertical skip> refers to one of the five glue-appending commands
just mentioned. The formal syntax for \<leaders> and for \<box or rule> is
\beginsyntax
<leaders>\is|\leaders|\alt|\cleaders|\alt|\xleaders|
<box or rule>\is<box>\alt<vertical rule>\alt<horizontal rule>
<vertical rule>\is|\vrule|<rule specification>
<horizontal rule>\is|\hrule|<rule specification>
<rule specification>\is<optional spaces>\alt<rule dimension><rule specification>
<rule dimension>\is[width]<dimen>\alt[height]<dimen>\alt[depth]<dimen>
\endsyntax
A glue item that produces ^{leaders} is appended to the current list.

\\^\<space token>.\enskip
Spaces have no effect in vertical modes.

\\^\<box>.\enskip
The box is constructed, and if the result is void nothing happens.
Otherwise the current vertical list receives (1)~interline glue, followed by
(2)~the new box, followed by (3)~vertical material that ^{migrates} out of the
new box (if the \<box> was an ^|\hbox| command). Then ^|\prevdepth| is
set to the new box's depth, and \TeX\ exercises the page builder.

\\^|\moveleft|\<dimen>\<box>, ^|\moveright|\<dimen>\<box>.\enskip
This acts exactly like an unadorned \<box> command, except that the new box
being appended to the vertical list is also shifted left or right by the
specified amount.

\\^|\unvbox|\<8-bit number>, ^|\unvcopy|\<8-bit number>.\enskip
If the specified box register is void, nothing happens. Otherwise that
register must contain a vbox. The vertical list inside that box is
appended to the current vertical list, without changing it in any way.
The value of\/ |\prevdepth| is not affected. The box register becomes void
after |\unvbox|, but it remains unchanged by |\unvcopy|.

\\\<horizontal rule>.\enskip
The specified ^{rule} is appended to the current list. Then |\prevdepth|
is set to $-1000\pt$; this will prohibit interline glue when the next box
is appended to the list.

\\^|\halign|\<box specification>|{|\<alignment material>|}|.\enskip
The ^\<alignment material> consists of a preamble followed by zero or more
lines to be aligned; see Chapter~22. \TeX\ enters a new level of grouping,
represented by the `|{|' and `|}|', within which changes to ^|\tabskip|
will be confined.  The alignment material can also contain optional
occurrences of `^|\noalign|\<filler>|{|\<vertical mode material>|}|'
between lines; this adds another level of grouping.  \TeX\ operates in
internal vertical mode while it works on the material in |\noalign| groups
and when it appends lines of the alignment; the resulting internal
vertical list will be appended to the enclosing vertical list after the
alignment is completed, and the page builder will be exercised. The value
of\/ |\prevdepth| at the time of the |\halign| is used at the beginning of
the internal vertical list, and the final value of\/ |\prevdepth| is carried
to the enclosing vertical list when the alignment is completed, so that the
interline glue is calculated properly at the beginning and end of the
alignment.  \TeX\ also enters an additional level of grouping when it
works on each individual entry of the alignment, during which time it acts
in restricted horizontal mode; the individual entries will be hboxed as
part of the final alignment, and their vertical material will ^{migrate}
to the enclosing vertical list.  The commands |\noalign|, ^|\omit|,
^|\span|, ^|\cr|, ^|\crcr|, and |&| (where |&| denotes an explicit or
implicit character of category~4) are intercepted by the alignment
process, en route to \TeX's stomach, so they will not appear as commands in
the stomach unless \TeX\ has lost track of what alignment they belong to.

\\^|\indent|.\enskip
The ^|\parskip| glue is appended to the current list, unless \TeX\ is
in internal vertical mode and the current list is empty.
Then \TeX\ enters unrestricted horizontal mode, starting the horizontal
list with an empty hbox whose width is ^|\parindent|. The ^|\everypar|
tokens are inserted into \TeX's input. The page builder is~exercised.
When the paragraph is eventually completed, horizontal mode will come to
an end as described in Chapter~25.

\\^|\noindent|.\enskip
This is exactly like |\indent|, except that \TeX\ starts out in horizontal
mode with an empty list instead of with an indentation.

\\^|\par|.\enskip
The primitive |\par| command has no effect when \TeX\ is in vertical mode,
except that the page builder is exercised in case something is present
on the contribution list, and the paragraph shape parameters are cleared.

\\|{|.\enskip
A character token of category 1, or a control sequence like~|\bgroup|
that has been |\let| equal to such a character token, causes \TeX\ to
start a new level of ^{grouping}. When such a group ends---with `|}|'---\TeX\
will undo the effects of non-global assignments without leaving whatever
mode it is in at that time.

\\Some commands are incompatible with vertical mode because they are
intrinsically horizontal. When the following commands appear in
vertical modes they cause \TeX\ to begin a ^{new paragraph}:
\beginsyntax
<horizontal command>\is<letter>\alt<otherchar>\alt^|\char|\alt<chardef token>
  \alt^|\noboundary|\alt^|\unhbox|\alt^|\unhcopy|\alt^|\valign|\alt^|\vrule|
  \alt^|\hskip|\alt^|\hfil|\alt^|\hfill|\alt^|\hss|\alt^|\hfilneg|
  \alt^|\accent|\alt^|\discretionary|\alt^|\-|\alt^|\|\]\alt|$|
\endsyntax
Here \<letter> and \<otherchar> stand for explicit or implicit character
tokens of categories 11 and~12. If any of these tokens occurs as a
command in vertical mode or internal vertical mode, \TeX\ automatically
performs an |\indent| command as explained above. This leads into
horizontal mode with the |\everypar| tokens in the input, after
which \TeX\ will see the \<horizontal command> again.

\\^|\end|.\enskip
This command is not allowed in internal vertical mode. In regular vertical
mode it terminates \TeX\ if the main vertical list is empty and
^|\deadcycles=0|. Otherwise \TeX\ backs up the |\end| command so that it can
be read again; then it exercises the page builder, after appending a
box/glue/penalty combination that will force the output routine to act.
\ (See the end of Chapter~23.)

\\^|\dump|.\enskip
(Allowed only in ^|INITEX|, not in production versions of \TeX.) \
This command is treated exactly like |\end|, but it must not appear
inside a group. It outputs a format file that can be loaded into \TeX's
memory at comparatively high speed to restore the current status.

\\None of the above: If any other primitive command of \TeX\ occurs
in vertical mode, an error message will be given, and \TeX\ will try
to recover in a reasonable way. For example, if a superscript or
subscript symbol appears, or if any other inherently mathematical
command is given, \TeX\ will try to insert a `|$|' (which will start a
paragraph and enter math mode). On the other hand if a totally
misplaced token like ^|\endcsname| or |\omit| or |\eqno| or |#| appears
in vertical mode, \TeX\ will simply ignore~it, after reporting
the error. You might enjoy trying to type some really stupid input,
just to see what happens. \ (Say `|\tracingall|' first, as explained
in Chapter~27, in order to get maximum information.)

\endchapter

The first and most striking feature is the Verticality of composition,
as opposed to the Horizontality of all anterior structural modes.
\author COCKBURN ^{MUIR}, {\sl Pagan or Christian?\/} (1860) % p61

\bigskip

Sometimes when I have finished a book I give a summary of the whole of it.
\author ROBERT WILLIAM ^{DALE}, {\sl Nine Lectures on Preaching} (1878)
 % viii.231

\eject
\beginchapter Chapter 25. Summary of\\Horizontal Mode

^^{horizontal mode}
Continuing the survey that was begun in Chapter 24, let us investigate
exactly what \TeX's digestive processes can do, when \TeX\ is building
lists in horizontal mode or in restricted horizontal mode.

\ninepoint
\def\\{\smallbreak\textindent{$\bull$}}
\medbreak
\centerline{$*\qquad*\qquad*$}
\medskip\noindent
Three asterisks, just like those that appear here, can be found near the
end of Chapter~24.
Everything preceding the three asterisks in that chapter applies to
horizontal mode as well as to vertical mode, so we need not repeat
all those rules. In particular, Chapter~24 explains assignment commands,
and it tells how kerns, penalties, marks, insertions, adjustments,
and ``whatsits'' are put into horizontal lists. Our present goal
is to consider the commands that have an intrinsically horizontal
flavor, in the sense that they behave differently in horizontal
mode than they do in vertical or math modes.

One of the things characteristic of horizontal mode is the ``^{space
factor},'' which modifies the width of spaces as described in Chapter~12.
If a command changes the value of\/ ^|\spacefactor|, that fact is
specifically noted here. The space factor is initially set to~1000, when
\TeX\ begins to form a horizontal list, except in the case of\/ |\valign| and
|\noalign| when the space factor of the outer list continues inside the
inner one.

\\^|\hskip|\<glue>, ^|\hfil|, ^|\hfill|, ^|\hss|, ^|\hfilneg|.\enskip
A glue item is appended to the current horizontal list.

\\\<leaders>\<box or rule>\<horizontal skip>.\enskip
Here ^\<horizontal skip> refers to one of the five glue-appending commands
just mentioned; the formal syntax for \<leaders> and for \<box or rule> is
given in Chapter~24. A glue item that produces ^{leaders} is appended.

\\^\<space token>.\enskip
Spaces append glue to the current list; the exact amount of glue depends on
|\spacefactor|, the current font, and the |\spaceskip| and |\xspaceskip|
parameters, as described in Chapter~12.

\\|\|\].\enskip ^^{control space}
A control-space command appends glue to the current list, using the same amount
that a \<space token> inserts when the space factor is 1000.

\\^\<box>.\enskip
The box is constructed, and if the result is void nothing happens.
Otherwise the new box is appended to the current list, and
the space factor is~set~to~1000.

\\^|\raise|\<dimen>\<box>, ^|\lower|\<dimen>\<box>.\enskip
This acts just like an unadorned \<box> command, except that the new box
being appended to the horizontal list is also shifted up or down by the
specified amount.

\\^|\unhbox|\<8-bit number>, ^|\unhcopy|\<8-bit number>.\enskip
If the specified box register is void, nothing happens. Otherwise that
register must contain an hbox. The horizontal list inside that box is
appended to the current horizontal list, without changing it in any way.
The value of\/ |\spacefactor| is not affected. The box register becomes void
after |\unhbox|, but it remains unchanged by |\unhcopy|.

\\\<vertical rule>.\enskip
The specified ^{rule} is appended to the current list, and the |\spacefactor|
is set to 1000.

\\^|\valign|\<box specification>|{|\<alignment material>|}|.\enskip
The ^\<alignment material> consists of a preamble followed by zero or more
columns to be aligned; see Chapter~22. \TeX\ enters a new level of grouping,
represented by the `|{|' and `|}|', within which changes to ^|\tabskip|
will be confined.  The alignment material can also contain optional
occurrences of `|\noalign|\<filler>|{|\<horizontal mode material>|}|'
between columns; this adds another level of grouping.  \TeX\ operates in
restricted horizontal mode while it works on the material in ^|\noalign|
\vadjust{\eject}% squeeze another line onto page 285, we need it on page 286!
groups and when it appends columns of the \hbox{alignment}; the resulting
internal horizontal list will be appended to the enclosing horizontal list
after the alignment is completed. The value of\/ |\spacefactor| at the time
of the |\valign| is used at the beginning of the internal horizontal list,
and the final value of\/ |\spacefactor| is carried to the enclosing
horizontal list when the alignment is completed.  The space factor is set
to 1000 after each column; hence it affects the results only in |\noalign|
groups.  \TeX\ also enters an additional level of grouping when it works
on each individual entry of the alignment, during which time it acts in
internal vertical mode; the individual entries will be vboxed as part of
the final alignment.
%No room for the following redundant remarks:
%  The commands |\noalign|, |\omit|, |\span|, |\cr|,
%|\crcr|, and |&| (where |&| denotes an explicit or implicit character of
%category~4) are intercepted by the alignment process, en route to \TeX's
%stomach, so they will not appear as commands in the stomach unless \TeX\
%has lost track of what alignment they belong to.

\\^|\indent|.\enskip
An empty box of width ^|\parindent| is appended to the current list, and
the space factor is set to 1000.

\\^|\noindent|.\enskip
This command has no effect in horizontal modes.

\\^|\par|.\enskip
The primitive |\par| command, also called ^|\endgraf| in plain \TeX,
does nothing in restricted horizontal mode. But it terminates horizontal
mode: The current list is finished off by doing ^|\unskip| ^|\penalty10000|
^^|\parfillskip|
|\hskip\parfillskip|, then it is broken into lines as explained in Chapter~14,
and \TeX\ returns to the enclosing vertical or internal vertical mode.
The lines of the paragraph are appended to the enclosing vertical list,
interspersed with interline glue and interline penalties, and with
the ^{migration} of vertical material that was in the horizontal list.
Then \TeX\ exercises the page builder.

\\|{|.\enskip
A character token of category 1, or a control sequence like~|\bgroup|
that has been |\let| equal to such a character token, causes \TeX\ to
start a new level of ^{grouping}. When such a group ends---with `|}|'---\TeX\
will undo the effects of non-global assignments without leaving whatever
mode it is in at that time.

\\Some commands are incompatible with horizontal mode because they are
intrinsically vertical. When the following commands appear in unrestricted
horizontal mode, they cause \TeX\ to conclude the current paragraph:
^^{paragraph end, implied}
\beginsyntax
<vertical command>\is^|\unvbox|\alt^|\unvcopy|\alt^|\halign|\alt^|\hrule|
  \alt^|\vskip|\alt^|\vfil|\alt^|\vfill|\alt^|\vss|\alt^|\vfilneg|%
  \alt^|\end|\alt^|\dump|
\endsyntax
The appearance of a \<vertical command> in restricted horizontal mode is
forbidden, but in regular horizontal mode it causes \TeX\ to insert the
token \cstok{par} into the input; after reading and expanding this \cstok{par}
token, \TeX\ will see the \<vertical command> token again. \ (The current
meaning of the control sequence ^|\par| will be used; \cstok{par} might no
longer stand for \TeX's |\par| primitive.)

\\\<letter>, \<otherchar>, \kern-1pt^|\char|\<8-bit number>, \<chardef token>,
\kern-1pt^|\noboundary|.\enskip
The most common commands of all are the character commands that tell
\TeX\ to append a character to the current horizontal
list, using the current font.
If two or more commands of this type occur in succession, \TeX\ processes
them all as a unit, converting to ligatures and/or
inserting kerns as directed by the font information. \ (Ligatures and
kerns may be influenced by invisible ``boundary'' characters at the left
and right, unless |\noboundary| appears.) \ Each character
command adjusts ^|\spacefactor|, using
the ^|\sfcode| table as described in Chapter~12.
In unrestricted horizontal mode, a
`|\discretionary{}{}{}|' item is appended after a character whose code is
the ^|\hyphenchar| of its font, or after a ligature formed from a sequence
that ends with such a character. ^^|\discretionary|

\\^|\accent|\<8-bit number>\<optional assignments>.
Here ^\<optional assignments> stands for zero or more \<assignment>
commands other than ^|\setbox|, possibly with \<filler>.
If the assignments are not followed by a \<character>, where
\<character> stands for any of the commands just discussed in the previous
paragraph, \TeX\ treats |\accent| as if it were |\char|, except that
the space factor is set to 1000. Otherwise the character that follows
the assignment is accented by the character that corresponds to the
\<8-bit number>. \ (The purpose of the intervening assignments is to
allow the accenter and accentee to be in different fonts.) \ If the
accent must be moved up or down, it is put into an hbox that is
raised or lowered. Then the accent is effectively superposed on the
character by means of kerns, in such a way that the width of the accent
does not influence the width of the resulting horizontal list.
Finally, \TeX\ sets |\spacefactor=1000|.

\\^|\/|.\enskip
If the last item on the current list is a character or ligature, an
explicit kern for its ^{italic correction} is appended.

\\^|\discretionary|\<disc text>\<disc text>\<disc text>.\enskip
A \<disc text> has the form `\<filler>|{|\<horizontal mode material>|}|',
where the material is processed in restricted horizontal mode and
should contain only fixed-width things.
More precisely, the horizontal list formed by each
^{\<disc text>} must consist only of characters, ligatures,
kerns, boxes, and rules; there should be no glue or penalty items, etc.
This command appends a discretionary item to the current list; see
Chapter~14 for the meaning of a discretionary item. The space factor is
not changed.

\\^|\-|.\enskip
This ``discretionary hyphen'' command is defined in Appendix H.

\\^|\setlanguage|\<number>.\enskip
See the conclusion of Appendix H.

\\|$|.\enskip
A ``^{math shift}'' character causes \TeX\ to enter math mode or display math
mode in the following way:
\TeX\ looks at the following token without expanding it. If that token
is a~|$| and if \TeX\ is currently in unrestricted horizontal mode,
then \TeX\ breaks the current paragraph into lines as explained above
(unless the current list is empty), returns to the enclosing vertical
mode or internal vertical mode, calculates values like |\prevgraf|
and |\displaywidth| and |\predisplaysize|, enters a new level of grouping,
inserts the |\everydisplay| tokens into the input, exercises the page
builder, processes `\<math mode material>|$$|' in display math mode, puts
the display into the enclosing vertical list as explained in Chapter~19
(letting vertical material ^{migrate}), exercises the page builder again,
increases |\prevgraf| by~3, and resumes horizontal mode again, with an
empty list and with the space factor equal to~1000. \ (You got that?) \
Otherwise \TeX\ puts the looked-at token back into the
input, enters a new level of grouping, inserts the |\everymath| tokens,
and processes `\<math mode material>|$|'; the math mode material
is converted to a horizontal list and appended to the current list,
surrounded by ``math-on'' and ``math-off'' items, and the space factor
is set to~1000. One consequence of these rules is that `|$$|' in
restricted horizontal mode simply yields an empty math formula.

\\None of the above: If any other primitive command of \TeX\ occurs
in horizontal mode, an error message will be given, and \TeX\ will try
to recover in a reasonable way. For example, if a superscript or
subscript symbol appears, or if any other inherently mathematical
command is given, \TeX\ will try to insert a `|$|' just before the
offending token; this will enter math mode.
%No room for the following redundant remarks:
%On the other hand if a totally
%misplaced token like |\endcsname| or |\omit| or |\eqno| or |#| appears
%in horizontal mode, \TeX\ will simply ignore it, after reporting
%the error. You might enjoy trying to type some really stupid input,
%just to see what happens. \ (Say `|\tracingall|' first, as explained
%in Chapter~27, in order to get maximum information.)

\endchapter

\strut{\rm Otherwise.} %
  You may reduce all\/ \kern.5pt{\rm Verticals} into\/ {\rm Horizontals}.
\author JOSEPH ^{MOXON}, {\sl A Tutor to Astronomie and Geographie\/} (1659)

\bigskip

\strut\tt!~You can't use `\bslash moveleft' in horizontal mode.
\author \TeX\ (1982)

\eject
\beginchapter Chapter 26. Summary of\\Math Mode

^^{math mode}
To conclude the survey that was begun in Chapter 24, let us investigate
exactly what \TeX's digestive processes can do when \TeX\ is building
lists in math mode or in display math mode.

\ninepoint
\def\\{\smallbreak\textindent{$\bull$}}
\medbreak
\centerline{$*\qquad*\qquad*$}
\medskip\noindent
Three asterisks, just like those that appear here, can be found near the
end of Chapter~24.
Everything preceding the three asterisks in that chapter applies to
math mode as well as to vertical mode, so we need not repeat
all those rules. In particular, Chapter~24 explains assignment commands,
and it tells how kerns, penalties, marks, insertions, adjustments,
and ``whatsits'' are put into math lists. Our present goal
is to consider the commands that have an intrinsically mathematical
flavor, in the sense that they behave differently in math
mode than they do in vertical or horizontal modes.

Math lists are somewhat different from \TeX's other lists because they
contain three-pronged ``^{atoms}'' (see Chapter~17). Atoms come in thirteen
flavors: Ord, Op, Bin, Rel, Open, Close, Punct, Inner, Over, Under, Acc,
Rad, and Vcent. Each atom contains three ``^{fields}'' called its ^{nucleus},
^{superscript}, and ^{subscript}; and each field is either empty or is filled
with a math symbol, a box, or a subsidiary math list. Math symbols, in turn,
have two components: a family number and a position number.

It's convenient to introduce a few more rules of syntax, in order to specify
what goes into a math list:
\beginsyntax
<character>\is<letter>\alt<otherchar>\alt^|\char|<8-bit number>\alt%
    <chardef token>
<math character>\is^|\mathchar|<15-bit number>\alt<mathchardef token>
  \alt|\delimiter|<27-bit number>
<math symbol>\is<character>\alt<math character>
<math field>\is<filler><math symbol>\alt<filler>|{|<math mode material>|}|
<delim>\is<filler>^|\delimiter|<27-bit number>
  \alt<filler><letter>\alt<filler><otherchar>
\endsyntax
We have already seen the concept of \<character> in Chapter~25.
Indeed, characters are \TeX's staple food: The vast majority of all
commands that reach \TeX's digestive processes in horizontal mode are instances
of the \<character> command, which specifies a number between 0 and~255 that
causes \TeX\ to typeset the corresponding character in the current font.
When \TeX\ is in math mode or display math mode, a \<character> command
takes on added significance: It specifies a number between 0 and~$32767=
2^{15}-1$. This is done by replacing the character number by its
^|\mathcode| value. If the
|\mathcode| value turns out to be $32768=\null$\hex{8000}, however,
the \<character>
is replaced by an ^{active character} token having the original character
code (0 to~255); \TeX\ forgets the original \<character> and expands this
active character according to the rules of Chapter~20.

A \<math character> defines a 15-bit number either by specifying it
directly with ^|\mathchar| or in a previous ^|\mathchardef|, or by
specifying a 27-bit |\delimiter| value; in the latter case, the least
significant 12~bits are discarded.

It follows that every \<math symbol>, as defined by the syntax above,
specifies a 15-bit number, i.e., a number between 0 and~32767.
Such a number can be represented in the form $4096c+256f+a$, where
$0\le c<8$, $0\le f<16$, and $0\le a<256$. If $c=7$, \TeX\ changes $c$~to~0;
and in this case if the current
value of\/ ^|\fam| is between 0 and~15, \TeX\ also replaces $f$~by~|\fam|.
This procedure yields, in all cases, a class number~$c$ between 0~and~6,
a family number~$f$ between 0~and~15, and a position number~$a$ between
0 and~255. \ (\TeX\ initializes the value of\/ |\fam| by implicitly
putting the assignment `|\fam=-1|' at the very beginning of\/ |\everymath|
and |\everydisplay|. Thus, the substitution of\/ |\fam| for~$f$ will occur
only if the user has explicitly changed~|\fam| within the formula.)

A \<math field> is used to specify the nucleus, superscript, or subscript
of an atom. When a \<math field> is a \<math symbol>, the $f$ and~$a$
numbers of that symbol go into the atomic field. Otherwise the \<math field>
begins with a~`|{|', which causes \TeX\ to enter a new level
of grouping and to begin a new math list; the ensuing \<math mode material>
is terminated by a~`|}|', at which point the group ends and the resulting
math list goes into the atomic field. If the math list turns out to be
simply a single Ord atom without subscripts or superscripts,
or an Acc whose nucleus is an Ord, the
enclosing braces are effectively removed.

A \<delim> is used to define both a ``small character'' $a$ in family~$f$
and a ``large character''~$b$ in family~$g$, where $0\le a,b\le255$
and $0\le f,g\le15$; these character codes are used to construct variable-size
^{delimiters}, as explained in Appendix~G\null. If the \<delim> is given
explicitly in terms of a 27-bit number, the desired codes are obtained
by interpreting that number as
$c\cdot2^{24}+f\cdot2^{20}+a\cdot2^{12}+g\cdot2^8+b$, ignoring the value
of~$c$. Otherwise the delimiter is specified as a \<letter> or
\<otherchar> token, and the 24-bit ^|\delcode| value of that character is
interpreted as $f\cdot2^{20}+a\cdot2^{12}+g\cdot2^8+b$.

\smallskip
Now let's study the individual commands as \TeX\ obeys them in math mode,
considering first the ones that have analogs in vertical and/or horizontal
mode:

\\^|\hskip|\<glue>, ^|\hfil|, ^|\hfill|, ^|\hss|, ^|\hfilneg|,
^|\mskip|\<muglue>.\enskip
A glue item is appended to the current math list.

\\\<leaders>\<box or rule>\<horizontal skip>.\enskip
Here ^\<horizontal skip> refers to one of the first five glue-appending
commands just mentioned; the formal syntax for \<leaders> and
for \<box or rule> is given in Chapter~24. A glue item that produces
^{leaders} is appended to the current list.

\\^|\nonscript|. A special glue item of width zero is appended; it will
have the effect of cancelling the following item on the list, if that item
is glue and if the |\nonscript| is eventually typeset in ``script style''
or in ``scriptscript style.''

\\^|\noboundary|. This command is redundant and therefore has no
effect; boundary ligatures are automatically disabled in math modes.

\\^\<space token>.\enskip
Spaces have no effect in math modes.

\\|\|\].\enskip ^^{control space}
A control-space command appends glue to the current list, using the same amount
that a \<space token> inserts in horizontal mode when the space factor is 1000.

\\^\<box>.\enskip
The box is constructed, and if the result is void nothing happens.
Otherwise a new Ord atom is appended to the current math list, and the
box becomes its nucleus.

\\^|\raise|\<dimen>\<box>, ^|\lower|\<dimen>\<box>.\enskip
This acts just like an unadorned \<box> command, except that the new box
being put into the nucleus is also shifted up or down by the specified amount.

\\^|\vcenter|\stretch\<box specification>\stretch
|{|\<vertical mode material>|}|.\enskip
A vbox is formed as if `|\vcenter|' had been `|\vbox|'. Then a new
^{Vcent} atom is appended to the current math list, and the box becomes
its nucleus.

\\\<vertical rule>.\enskip
A ^{rule} is appended to the current list (not as an atom).

\\^|\halign|\<box specification>|{|\<alignment material>|}|.\enskip
This command is allowed only in display math mode, and only when the current
math list is empty. The alignment is carried out exactly as if it were done in
the enclosing vertical mode (see Chapter~24), except that the lines are shifted
right by the ^|\displayindent|. The closing `|}|' may be followed by
optional \<assignment> commands other than ^|\setbox|,
after which `|$$|'~must conclude
the display. \TeX\ will insert the ^|\abovedisplayskip| and
^|\belowdisplayskip| glue before and after the result of the alignment.

\\^|\indent|.\enskip
An empty box of width ^|\parindent| is appended to the current list, as the
nucleus of a new Ord atom.

\\^|\noindent|.\enskip
This command has no effect in math modes.

\\|{|\<math mode material>|}|.\enskip
A character token of category 1, or a control sequence like~|\bgroup|
that has been |\let| equal to such a character token, causes \TeX\ to
start a new level of ^{grouping} and also to begin work on a new
math list. When such a group ends---with `|}|'---\TeX\
uses the resulting math list as the nucleus of a new Ord atom that
is appended to the current list. If the resulting math list is a
single Acc atom, however (i.e., an accented quantity),
that atom itself is appended.

\\\<math symbol>.\enskip
(This is the most common command in math mode; see the syntax near the
beginning of this chapter.) \ A math symbol determines three values,
$c$,~$f$, and~$a$, as explained earlier. \TeX\ appends an atom to the current
list, where the atom is of type Ord, Op, Bin, Rel, Open, Close, or Punct,
according as the value of~$c$ is 0,~1, 2, 3, 4, 5, or~6. The nucleus of
this atom is the math symbol defined by $f$ and~$a$.

\\^\<math atom>\<math field>.\enskip
A \<math atom> command is any of the following:
\begindisplay
|\mathord|\alt|\mathop|\alt|\mathbin|\alt|\mathrel|\alt|\mathopen|\cr
\qquad\alt|\mathclose|\alt|\mathpunct|\alt|\mathinner|%
  \alt|\underline|\alt|\overline|\cr
\enddisplay
\TeX\ processes the \<math field>, then appends a new atom of the
specified type to the current list; the nucleus of this atom contains
the specified field.

\\^|\mathaccent|\<15-bit number>\<math field>.\enskip
\TeX\ converts the \<15-bit number> into $c$, $f$, and~$a$ as it does
with any |\mathchar|. Then it processes the \<math field> and appends a
new Acc atom to the current list. The nucleus of this atom contains
the specified field; the accent character in this atom contains $(a,f)$.

\\^|\radical|\<27-bit number>\<math field>.\enskip
\TeX\ converts the \<27-bit number> into $a$, $f$, $b$, and~$g$ as it does
with any |\delimiter|. Then it processes the \<math field> and appends a
new Rad atom to the current list. The nucleus of this atom contains
the specified field; the delimiter field in this atom contains
$(a,f)$ and $(b,g)$.

\\^\<superscript>\<math field>.\enskip
A \<superscript> command is an explicit or implicit character token of
category~7.  If the current list does not end with an atom, a new Ord atom
with all fields empty is appended; thus the current list will end with an
atom, in~all cases.  The superscript field of this atom should be empty; it
is made nonempty by changing it to the result of the specified \<math field>.

\\^\<subscript>\<math field>.\enskip
A \<subscript> command is an explicit or implicit character token of
category~8.  It acts just like a \<superscript> command, except, of course,
that it affects the subscript field instead of the superscript field.

\\^|\displaylimits|, ^|\limits|, ^|\nolimits|.\enskip These commands are
allowed only if the current list ends with an Op atom. They modify a
special field in that Op atom, specifying what conventions should be used
with respect to limits.  The normal value of that field is |\displaylimits|.

\\^|\/|.\enskip
A kern of width zero is appended to the current list. \ (This will have the
effect of adding the italic correction to the previous character, if the
italic correction wouldn't normally have been added.)

\\^|\discretionary|\<disc text>\<disc text>\<disc text>.
This command has the same effect as in horizontal mode (see Chapter~25), but the
third ^{\<disc text>} must produce an empty list.

\\^|\-|.\enskip
This command is usually equivalent to `|\discretionary{-}{}{}|';
the `|-|' is therefore interpreted as a ^{hyphen}, not as a minus sign.
\ (See Appendix~H.)

\def\s{\hskip0pt plus1pt}
\\^|\mathchoice|\s
$\langle$filler$\rangle$\s|{|\s$\langle$math mode material$\rangle$\s|}|\s
$\langle$filler$\rangle$\s|{|\s$\langle$math mode material$\rangle$\s|}|\break
$\langle$filler$\rangle$|{|$\langle$math mode material$\rangle$|}|
$\langle$filler$\rangle$|{|$\langle$math mode material$\rangle$|}|.
Four math lists, which are defined as in the
second alternative of a \<math field>, are
recorded in a ``choice item'' that is appended to the current list.

\\^|\displaystyle|, ^|\textstyle|, ^|\scriptstyle|,
^|\scriptscriptstyle|.\enskip A style-change item that corresponds to the
specified style is appended to the current list.

\\^|\left|\<delim>\<math mode material>^|\right|\<delim>.\enskip
\TeX\ begins a new group, and processes the \<math mode material>
by starting out with a new math list that begins with a left boundary
item containing the first delimiter. This group must be terminated by
`|\right|', at which time the internal math list is completed with a
right boundary item containing the second delimiter. Then \TeX\
appends an Inner atom to the current list; the nucleus of this atom
contains the internal math list.

\\\<generalized fraction command>.\enskip
This command takes one of six forms:
\begindisplay
^|\over|\alt^|\atop|\alt^|\above|\<dimen>\cr
\qquad\alt^|\overwithdelims|\<delim>\<delim>\cr
\qquad\alt^|\atopwithdelims|\<delim>\<delim>\cr
\qquad\alt^|\abovewithdelims|\<delim>\<delim>\<dimen>\cr
\enddisplay
(See Chapter 17.) \ When \TeX\ sees a \<generalized fraction command> it
takes the entire current list and puts it into the numerator field of a
generalized fraction item. The denominator field of this new item is
temporarily empty; the left and right delimiter fields are set equal
to the specified delimiter codes. \TeX\ saves this generalized fraction
item in a special place associated with the current level of math mode
processing. \ (There should be no other generalized fraction item in
that special place, because constructions like `|a\over b\over c|' are
illegal.) \ Then \TeX\ makes the current list empty and continues to
process commands in math mode. Later on, when the current level of math
mode is completed (either by coming to a~`|$|' or a~`|}|' or a~|\right|,
depending on the nature of the current group), the current list will be
moved into the denominator field of the generalized fraction item that
was saved; then that item, all by itself, will take the place of the
entire list. However, in the special case that the current list began
with |\left| and will end with |\right|, the boundary items will be
extracted from the numerator and denominator of the generalized fraction,
and the final list will consist of three items: left boundary,
generalized fraction, right boundary. \ (If you want to watch the process
by which math lists are built, you might find it helpful to type `^|\showlists|'
while \TeX\ is processing the denominator of a generalized fraction.)

\\^\<eqno>\<math mode material>|$|.\enskip
Here \<eqno> stands for either ^|\eqno| or ^|\leqno|; these commands
are allowed only in display math mode. Upon reading \<eqno>, \TeX\
enters a new level of grouping, inserts the ^|\everymath| tokens,
and enters non-display math mode to put the \<math mode material>
into a math list. When that math list is completed, \TeX\ converts it
to a horizontal list and puts the result into a box that will be used
as the equation number of the current display. The closing |$|~token
will be put back into the input, where it will terminate the display.

\\|$|.\enskip
If \TeX\ is in display math mode, it reads one more token, which must
also be~|$|. In either case, the math-shift command terminates the current
level of math mode processing and ends the current group, which should
have begun with either~|$| or~\<eqno>.  Once the math list is finished, it
is converted into a horizontal list as explained in Appendix~G\null. \TeX\
scans \<one optional space> after completing a displayed formula; this is
usually the implicit space at the end of a line in the input file.

\\^|\unhbox|\<8-bit number>, ^|\unhcopy|\<8-bit number>.\enskip
The specified box register must be void. Nothing happens.

\\None of the above: If any other primitive command of \TeX\ occurs in
math mode, an error message will be given, and \TeX\ will try to recover
in a reasonable way. For example, if a |\par| command appears, or if any
other inherently non-mathematical command is given, \TeX\ will try to
insert a `|$|' just before the offending token; this will lead out of math
mode. On the other hand if a totally misplaced token like |\endcsname| or
|\omit| or |#| appears in math mode, \TeX\ will simply ignore it, after
reporting the error. You might enjoy trying to type some really stupid
input, just to see what happens. \ (Say `|\tracingall|' first, as
explained in Chapter~27, in order to get maximum information.)

\ddangerexercise ^{Powers of ten}: The whole \TeX\ language has now been
summarized completely. To~demonstrate how much you know, name all of the ways
you can think of in which the numbers 10, ^^{Derek, Bo} 100, 1000, 10000,
and 100000 have special significance to \TeX.
\answer Radix 10 notation is used for numeric constants and for the output
of numeric data. The first 10 |\count| registers are displayed at each
|\shipout|, and their values are recorded on the |dvi| file at such times.
% Also, TeX was first implemented on a DEC-10.
% The \catcode for <space> is 10.
% My birthday is January 10.
% Can this all be just a coincidence?
A box whose glue has stretched or shrunk to its stated stretchability
or shrinkability has badness 100; this badness value separates ``loose''
boxes from ``very loose'' or ``underfull'' ones. \TeX\ will scroll up
to 100 errors in a single paragraph before giving up (see Chapter~27).
The normal values of\/ |\spacefactor| and |\mag| are 1000. A~|\prevdepth|
value of $-1000\pt$ suppresses interline glue. The badness rating of a
box is at most 10000, except that the |\badness| of
an overfull box is 1000000.  |INITEX| initializes |\tolerance| to
10000, thereby making all line breaks feasible. Penalties of 10000 or more
prohibit breaks; penalties of $-10000$ or less make breaks mandatory. The
cost of a page break is 100000, if the badness is 10000 and if the
associated penalties are less than 10000 in magnitude (see Chapter~15).

\ddangerexercise ^{Powers of two}: Name all of the ways
you can think of in which the numbers 8, 16, 32, 64, 128, 256, \dots\
have special significance to \TeX.
\answer \TeX\ allows constants to be expressed in radix 8 (octal)
or radix~16 (hexadecimal) notation, and it uses hexadecimal notation to
display |\char| and |\mathchar| codes. There are 16 families for math
fonts, 16 input streams for |\read|, 16 output streams for |\write|.
A |\catcode| value must be less than~16. The notation |^^?|, |^^@|,
|^^A| specifies characters whose codes differ by~64
from the codes of |?|, |@|, |A|; this convention applies only to
characters with ASCII codes less than~128. There are 256 possible characters,
hence 256 entries in each of the |\catcode|, |\mathcode|,
|\lccode|, |\uccode|, |\sfcode|, and |\delcode| tables. All
|\lccode|, |\uccode|, and |\char| values
must be less than~256. A font has at most 256 characters. There are
256~|\box| registers, 256~|\count| registers, 256~|\dimen| registers,
256~|\skip| registers, 256~|\muskip| registers, 256~|\toks| registers,
256~hyphenation tables.
The ``at size'' of a font must be less than~$2048\pt$, i.e.,~$2^{11}\pt$.
Math delimiters are encoded by multiplying the math~code of the ``small
character'' by~$2^{12}$. The magnitude of
a~\<dimen> value must be less than~$16384\pt$, i.e.,~$2^{14}\pt$;
similarly, the \<factor> in a~\<fil dimen> must be less than~$2^{14}$.
A~|\mathchar| or |\spacefactor| or |\sfcode| value must be less than~$2^{15}$;
a~|\mathcode| or |\mag| value must be less than or equal to~$2^{15}$,
and $2^{15}$ denotes an ``active'' math character. There
are $2^{16}\rm\,sp$ per~pt. A~|\delcode| value
must be less than~$2^{24}$; a~|\delimiter|, less than $2^{27}$.
The |\end| command sometimes contributes
a penalty of $-2^{30}$ to the current page. A~\<dimen> must be less than
$2^{30}\rm\,sp$ in absolute value; a~\<number> must be
less than $2^{31}$ in absolute value.

\endchapter

Mathematics is known in the trade as {\rm difficult,} or {\rm penalty, copy}
because it is slower, more difficult, and more expensive to set in type
than any other kind of copy normally occurring in books and journals.
\author UNIVERSITY OF ^{CHICAGO} PRESS, {\sl A Manual of Style\/} %
  (1969) % 12th edition, page 295

\bigskip

The tale of Math is a complex one,
and it resists both a simple plot summary
and a concise statement of its meaning.
\author PATRICK K. ^{FORD}, {\sl The Mabinogi\/} (1977) % p89
   % from his introduction to "Math Son of Mathonwy"

\eject
\beginchapter Chapter 27. Recovery from\\Errors

OK, everything you need to know about \TeX\ has been explained---unless you
happen to be fallible. If you don't plan to make any errors, don't bother to
read this chapter. Otherwise you might find it helpful to make use of some
of the ways that \TeX\ tries to pinpoint bugs in your manuscript.

In the trial runs you did when reading Chapter 6, you learned the general
form of ^{error messages}, and you also learned the various ways in which
you can respond to \TeX's complaints. With practice, you will be able to
correct most errors ``online,'' as soon as \TeX\ has detected them, by
inserting and deleting a few things. The right way to go about this is to
be in a mellow mood when you approach \TeX, and to regard the error
messages that you get as amusing puzzles---``Why did the machine do
that?''---rather than as personal insults.

\TeX\ knows how to issue more than a hundred different sorts of error messages,
and you probably never will encounter all of them, because some types of
mistakes are very hard to make. We discussed the ``undefined control sequence''
error in Chapter~6; let's take a look at a few of the others now.

If you misspell the name of some unit of measure---for example, if you
type `|\hsize=4im|' instead of `|\hsize=4in|'---you'll get an error
message that looks something like this:
\begintt
! Illegal unit of measure (pt inserted).
<to be read again>
                   i
<to be read again>
                   m
<*> \hsize=4im
               \input story
?
\endtt
^^|Illegal unit|
\TeX\ needs to see a legal unit before it can proceed; so in this case it has
implicitly inserted `|pt|' at the current place in the input, and it has
set |\hsize=4pt|.

What's the best way to ^{recover} from such an error? Well, you should
always type `|H|' or `|h|' to see the help message, if you aren't sure
what the error message means. Then you can look at the lines of context
and see that \TeX\ will read `|i|' and then `|m|' and then `~|\input|
|story|~' if you simply hit \<return> and carry on. Unfortunately, this
easy solution isn't very good, because the `|i|' and~`|m|' will be typeset
as part of the text of a new paragraph.  A much more graceful recovery is
possible in this case, by first typing~`|2|'.  This tells \TeX\ to discard
the next two tokens that it reads; and after \TeX\ has done so, it will
stop again in order to give you a chance to look over the new situation.
Here is what you will see:
\begintt
<recently read> m
|indent
<*> \hsize=4im
               \input story
?
\endtt
Good; the `|i|' and `|m|' are read and gone. But if you hit \<return> now,
\TeX\ will `|\input story|' and try to typeset the |story.tex| file with
|\hsize=4pt|; that won't be an especially exciting experiment, because it
will simply produce dozens of overfull boxes, one for every syllable
of the story. Once again there's a better way: You can insert the command
that you had originally intended, by typing
\begintt
I\hsize=4in
\endtt
now. This instructs \TeX\ to change |\hsize| to the correct value, after
which it will |\input story| and you'll be on your way.

\exercise Ben ^{User} typed `|8|', not `|2|', in response to the
error message just considered; his idea was to delete `|i|', `|m|',
`|\input|', and the five letters of `|story|'. But \TeX's response was
\begintt
<*> \hsize=4im \input stor
                          y
\endtt
Explain what happened.
\answer He forgot to count the space; \TeX\ deleted `|i|', `|m|',
`\]', `|\input|', and four letters. (But all is not lost; he can
type `|1|' or `|2|', then \<return>, and after being prompted by `|*|' he
can enter a new line of input.)

\TeX\ usually tries to recover from errors either by ignoring a command
that it doesn't understand, or by inserting something that will keep it
happy. For example, we saw in Chapter~6 that \TeX\ ignores an
undefined control sequence; and we just observed that \TeX\ inserts
`|pt|' when it needs a physical unit of measure.
Here's another example where \TeX\ puts something in:
\begintt
! Missing $ inserted.
<inserted text>
                $
<to be read again>
                   ^
l.11 the fact that 32768=2^
                           {15} wasn't interesting
? H
I've inserted a begin-math/end-math symbol since I think
you left one out. Proceed, with fingers crossed.
\endtt
^^{help message} ^^|Missing|
(The user has forgotten to enclose a formula in |$| signs, and \TeX\
has tried to recover by inserting one.) \ In this case the \<inserted text>
is explicitly shown, and it has not yet been read; by contrast, our
previous example illustrated a case where \TeX\ had already internalized
the `|pt|' that it had inserted. Thus the user has a chance here to
remove the inserted~`|$|' before \TeX\ really sees it.

What should be done? The error in this example occurred before \TeX\ noticed
anything wrong; the characters `32768=2' have already been typeset in
horizontal mode. There's no way to go back and cancel the past, so
the lack of proper spacing around the `=' cannot be fixed. Our goal of
error recovery in this case is therefore not to produce perfect
output; we want rather to proceed in some way so that \TeX\ will pass by
the present error and detect subsequent ones. If we were simply to hit
\<return> now, our aim would not be achieved, because \TeX\ would
typeset the ensuing text as a math formula: `$^{15} wasn't interesting
\ldots$'\thinspace; another error would be detected when the paragraph is
found to end before any closing~`|$|' has appeared. On the other hand,
there's a more elaborate way to recover, namely to type~`|6|' and then
`|I$^{15}$|'; this deletes `|$^{15}|' and inserts a
correct partial formula. But that's more complicated than necessary.
The best solution in this case is to type just~`|2|' and then go on;
\TeX\ will typeset the incorrect equation `32768=215', but the important
thing is that you will be able to check out the rest of the document as if
this error hadn't occurred.

\dangerexercise Here's a case in which a backslash was inadvertently omitted:
\begintt
! Missing control sequence inserted.
<inserted text>
                \inaccessible
<to be read again>
                   m
l.10 \def m
           acro{replacement}
\endtt
% The help message issued by TeX refers to "exercise 27.2 in The TeXbook"!
\TeX\ needs to see a control sequence after `|\def|', so it has inserted one
that will allow the processing to continue. \ (This control sequence is
shown as `^|\inaccessible|', but it has no relation to any control
sequence that you can actually specify in an error-free manuscript.) \
If you simply hit \<return> at this point, \TeX\ will define the
inaccessible control sequence, but that won't do you much good;
later references to |\macro| will be undefined. Explain how to
recover from this error so that the effect will be the same as if
line~10 of the input file had said `|\def\macro{replacement}|'.
\answer First delete the unwanted tokens, then insert what you want:
Type `|6|' and then `|I\macro|'. \ (Incidentally, there's a sneaky
way to get at the |\inaccessible| control sequence by typing
\begintt
I\garbage{}\let\accessible=
\endtt
in response to an error message like this. The author ^^{Knuth} designed
\TeX\ in such a way that you can't destroy anything by playing such
nasty tricks.)

\dangerexercise When you use the `|I|' option to respond to an error message,
the rules of Chapter~8 imply that \TeX\ removes all spaces from the
right-hand end of the line. Explain how you can use the `|I|'~option
to insert a ^{space}, in spite of this fact.
\answer `|I|\]|%|' does the trick, if |%| is a ^{comment character}.

Some of the toughest errors to deal with are those in which you make a
mistake on line~20 (say), but \TeX\ cannot tell that anything is amiss
until it reaches line~25 or so. For example, if you forget a `|}|' that
completes the argument to some macro, \TeX\ won't notice any problem
until reaching the end of the next paragraph. In such cases you probably
have lost the whole paragraph; but \TeX\ will usually be able to get
straightened out in time to do the subsequent paragraphs as if
nothing had happened. A ``^{runaway argument}'' will be displayed,
and by looking at the beginning of that text you should be able to
figure out where the missing `|}|' belongs.

It's wise to remember that the first error in your document may well spawn
spurious ``errors'' later on, because anomalous commands can inflict
serious injury on \TeX's ability to cope with the subsequent material.
But most of the time you will find that a single run through the
machine will locate all of the places in which your input conflicts
with \TeX's rules.

When your error is due to misunderstanding rather than mistyping,
the situation is even more serious: \TeX's error messages will probably
not be very helpful, even if you ask \TeX\ for help.
If you have unknowingly redefined an important control sequence---for
example, if you have said `|\def\box{...}|'---all sorts of strange
disasters might occur. Computers
aren't clairvoyant, and \TeX\ can only explain what looks wrong from
its own viewpoint; such an explanation is bound to be mysterious
unless you can understand the machine's attitude. The solution
to this problem is, of course, to seek human counsel and advice;
or, as a last resort, to read the instructions in Chapters 2, 3, \dots, 26.

\newcount\serialnumber
\def\firstnumber{\serialnumber=0 }
\def\nextnumber{\advance \serialnumber by 1
  \number\serialnumber)\nobreak\hskip.2em }
\dangerexercise J. H. ^{Quick} (a student) once defined the following
set of macros:
\begintt
\newcount\serialnumber
\def\firstnumber{\serialnumber=0 }
\def\nextnumber{\advance \serialnumber by 1
  \number\serialnumber)\nobreak\hskip.2em }
\endtt
Thus he could type, for example,
\begintt
\firstnumber
\nextnumber xx, \nextnumber yy, and \nextnumber zz
\endtt
and \TeX\ would typeset
\firstnumber
`\nextnumber xx, \nextnumber yy, and \nextnumber zz'.
Well, this worked fine, and he showed the macros to his buddies.
But several months later he received a frantic phone call; one of his
friends had just encountered a really ^{weird error} message:
\begintt
! Missing number, treated as zero.
<to be read again>
                   c
l.107 \nextnumber minusc
                        ule chances of error
?
\endtt
Explain what happened, and advise Quick what to do.
\answer The `^|minus|' of `|minuscule|' was treated as part of
the |\hskip| command in |\nextnumber|. Quick should put `|\relax|'
at the end of his macro. \ (The ^{keywords} ^|l|, ^|plus|, |minus|,
^|width|, ^|depth|, or ^|height| might just happen to occur in text
when \TeX\ is reading a glue specification or a rule specification;
designers of general-purpose macros should guard against this.
If you get a `^|Missing number|' error and you can't guess
why \TeX\ is looking for a number, plant the instruction
`|\tracingcommands=1|' shortly before the error point; your log file
will show what command \TeX\ is working~on.)

Sooner or later---hopefully sooner---you'll get \TeX\ to process your
whole file without stopping once to complain. But maybe the output
still won't be right; the mere fact that \TeX\ didn't stop doesn't mean
that you can avoid proofreading. At this stage it's usually easy to
see how to fix typographic errors by correcting the input. Errors of
layout can be overcome by using methods we have discussed before:
Overfull boxes can be cured as described in Chapter~6; bad breaks can
be avoided by using ties or |\hbox| commands as discussed in Chapter~14;
math formulas can be improved by applying the principles of
Chapters 16--19.

But your output may contain seemingly inexplicable errors. For example,
if you have specified a font at some magnification that is not supported
by your printing software, \TeX\ will not know that there is any
problem, but the program that converts your |dvi| file to hardcopy
might not tell you that it has substituted an ``approximate'' font
for the real one; the resultant spacing may look quite horrible.

If you can't find out what went wrong, try the old trick of simplifying
your program: Remove all the things that do work, until you obtain
the shortest possible input file that fails in the same way as the
original. The shorter the file, the easier it will be for you or somebody
else to pinpoint the problem.

Perhaps you'll wonder why \TeX\ didn't put a blank space in some
position where you think you typed a space. Remember that \TeX\ ignores
^{spaces} that follow control words, when it reads your file.
\ (\TeX\ also ignores a space after a \<number> or a \<unit of
measure> that appears as an argument to a primitive command; but if
you are using properly designed macros, such rules will not concern
you, because you will probably not be using primitive commands~directly.)

\ddanger On the other hand, if you are designing macros, the task of
troubleshooting can be a lot more complicated. For example, you may
discover that \TeX\ has emitted three blank spaces when it processed some
long sequence of complicated code, consisting of several dozen commands.
How can you find out where those spaces crept in? The answer is to set
`^|\tracingcommands||=1|', as mentioned in Chapter~13. This tells \TeX\ to
put an entry in your log file whenever it begins to execute a primitive
command; you'll be able to see when the command is `|blank| |space|'.

\danger Most implementations of \TeX\ allow you to ^{interrupt} the program
in some way. This makes it possible to diagnose the causes of ^{infinite
loops}. \TeX\ switches to ^|\errorstopmode| when interrupted; hence
you have a chance to insert commands into the input: You can abort the
run, or you can ^|\show| or change the current contents of control sequences,
registers, etc. You can also get a feeling for where \TeX\ is spending most
of its time, if you happen to be using an inefficient macro, since random
interrupts will tend to occur in whatever place \TeX\ visits most often.

\danger Sometimes an error is so bad that \TeX\ is forced to quit
prematurely. For example, if you are running in ^|\batchmode| or
^|\nonstopmode|, \TeX\ makes an ``^{emergency stop}'' if it needs
input from the terminal; this happens when a necessary file cannot
be opened, or when no ^|\end| command was found in the input
document. Here are some of the messages you might get just before
\TeX\ gives up the ghost: \enddanger

{\ninepoint
\def\fatal#1. {\medbreak{\tt#1.}\par\nobreak\smallskip\noindent\ignorespaces}
\fatal
Fatal format file error; I'm stymied.
^^|Fatal format file error|
This means that the preloaded format you have specified cannot be used,
because it is corrupted or was prepared for a different version of \TeX.
\fatal
That makes 100 errors; please try again.
\TeX\ has scrolled past 100 errors since the last paragraph ended, so
it's probably in an~endless ^{loop}. ^^{infinite loop}
\fatal
Interwoven alignment preambles are not allowed.
^^|Interwoven alignment preambles|
If you have been so devious as to get this message, you will understand
it, and you will deserve no sympathy.
\fatal
I can't go on meeting you like this.
^^|I can't go on|
A previous error has gotten \TeX\ out of whack. Fix it and try again.
\fatal
This can't happen.
^^|This can't happen|
Something is wrong with the \TeX\ you are using. Complain fiercely.
\goodbreak}

\danger There's also a dreadful message that \TeX\ issues only with
great reluctance. But it can happen:
\begintt
TeX capacity exceeded, sorry.
\endtt
^^|TeX capacity exceeded|
This, alas, means that you have tried to stretch \TeX\ too far. The
message will tell you what part of \TeX's memory has become overloaded;
one of the following fourteen things will be mentioned:
\begindisplay
|number of strings|\qquad(names of control sequences and files)\cr
|pool size|\qquad(the characters in such names)\cr
|main memory size|\qquad(boxes, glue, breakpoints, token lists,
        characters, etc.)\cr
|hash size|\qquad(control sequence names)\cr
|font memory|\qquad(font metric data)\cr
|exception dictionary|\qquad(hyphenation exceptions)\cr
|input stack size|\qquad(simultaneous input sources)\cr
|semantic nest size|\qquad(unfinished lists being constructed)\cr
|parameter stack size|\qquad(macro parameters)\cr
|buffer size|\qquad(characters in lines being read from files)\cr
|save size|\qquad(values to restore at group ends)\cr
|text input levels|\qquad(|\input| files and error insertions)\cr
|grouping levels|\qquad(unfinished groups)\cr
|pattern memory|\qquad(hyphenation pattern data)\cr
\enddisplay
The current amount of memory available will also be shown.

\danger If you have a job that doesn't overflow \TeX's capacity, yet
you want to see just how closely you have approached the limits,
just set ^|\tracingstats| to a positive value before the end of your
job. The log file will then conclude with a report on your actual
usage of the first eleven things named above (i.e., the number of strings,
\dots, the save size), in that order. ^^{stack positions}
Furthermore, if you set |\tracingstats| equal to 2~or~more, \TeX\
will show its current memory usage whenever it
does a ^|\shipout| command. Such statistics are broken into two
parts; `|490&5950|' means, for example, that 490 words are being used
for ``large'' things like boxes, glue, and
breakpoints, while 5950 words are being used for ``small'' things like
tokens and characters.

\danger What can be done if \TeX's capacity is exceeded? All of the
above-listed components of the capacity can be increased, provided
that your computer is large enough; in fact, the space necessary to
increase one component can usually be obtained by decreasing some
other component, without increasing the total size of \TeX\null.
If you have an especially important application, you may be able
to convince your local system people to provide you with a special
\TeX\ whose capacities have been hand-tailored to your needs.
But before taking such a drastic step, be sure that you are using
\TeX\ properly. If you have specified a gigantic paragraph or
a gigantic alignment that spans more than one page, you should
change your approach, because \TeX\ has to read all the way to the
end before it can complete the line-breaking or the alignment
calculations; this consumes huge amounts of memory space. If you have
built up an enormous macro library, you should remember that \TeX\
has to remember all of the replacement texts that you define; therefore
if memory space is in short supply, you should load only the macros
that you need. \ (See Appendices B and~D\null, for ideas on how to make
macros more compact.)

\danger Some erroneous \TeX\ programs will overflow any finite
memory capacity. For example, after `|\def\recurse{(\recurse)}|', the
^^{recursion} use of\/ |\recurse| will immediately bomb out:
\begintt
! TeX capacity exceeded, sorry [input stack size=80].
\recurse ->(\recurse
                     )
\recurse ->(\recurse
                     )
...
\endtt
The same sort of error will obviously occur no matter how much you increase
\TeX's input stack size.

\ddanger The special case of ``^|save size|'' capacity exceeded is one
of the most troublesome errors to correct, especially if you run into
the error only on long jobs. \TeX\ generally uses up two words of save
size whenever it performs a non-global assignment to some quantity
whose previous value was not assigned at the same level of ^{grouping}.
When macros are written properly, there will rarely be a need for
more than 100 or~so things on the ``^{save stack}''; but it's possible to
make save stack usage grow without limit if you make both local and
^{global assignments} to the same variable. You can figure out what \TeX\
puts on the save stack by setting ^|\tracingrestores||=1|; then your
log file will record information about whatever is removed from the
stack at the end of a group. For example, let |\a| stand for the
command `|\advance\day|~|by|~|1|'; let |\g| stand for
`|\global\advance\day|~|by|~|1|'; and consider the following commands:
\begintt
\day=1 {\a\g\a\g\a}
\endtt
The first |\a| sets |\day=2| and remembers the old value |\day=1| by putting
it on the save stack. The first |\g| sets |\day=3|, globally; nothing needs
to go on the save stack at the time of a global assignment. The next |\a|
sets |\day=4| and remembers the old value |\day=3| on the save stack.
Then |\g| sets |\day=5|; then |\a| sets |\day=6| and remembers |\day=5|.
^^{right brace}
Finally the `|}|' causes \TeX\ to go back through the save stack; if
|\tracingrestores=1| at this point, the log file will get the following
data:
\begintt
{restoring \day=5}
{retaining \day=5}
{retaining \day=5}
\endtt
Explanation: The |\day| parameter is first restored to its global value~5.
Since this value is global, it will be retained, so the other saved values
(|\day=3| and |\day=1|) are essentially ignored. Moral: If you find \TeX\
retaining a lot of values, you have a set of macros that could cause
the save stack to overflow in large enough jobs. To prevent this, it's
usually wise to be consistent in your assignments to each variable that
you use; the assignments should either be global always or local always.

\ddanger \TeX\ provides several other kinds of tracing in addition to
|\tracingstats| and |\tracingrestores|: We have already discussed
|\tracingcommands| in Chapters 13 and~20, |\tracingparagraphs| in
Chapter~14, |\tracingpages| in Chapter~15, and |\tracingmacros| in
Chapter~20. There is also ^|\tracinglostchars|, which (if positive)
causes \TeX\ to record each time a character has been dropped because
it does not appear in the current font; and ^|\tracingoutput|, which
(if positive) causes \TeX\ to display in symbolic form the contents
of every box that is being shipped out to the ^|dvi|~file. ^^|\shipout|
The latter allows you to see if things have been typeset properly, if
you're trying to decide whether some anomaly was caused by \TeX\ or
by some other software that acts on \TeX's output.

\danger When \TeX\ displays a box as part of diagnostic output, the amount
of data is controlled by two parameters called ^|\showboxbreadth| and
^|\showboxdepth|. The first of these, which plain \TeX\ sets equal to~5,
tells the maximum number of items shown per level; the second, which plain
\TeX\ sets to~3, tells the deepest level. For example, a small box
whose full contents are ^^{internal box format} ^^{symbolic box format}
\begintt
\hbox(4.30554+1.94444)x21.0, glue set 0.5
.\hbox(4.30554+1.94444)x5.0
..\tenrm g
.\glue 5.0 plus 2.0
.\tenrm || (ligature ---)
\endtt
will be abbreviated as follows when |\showboxbreadth=1| and |\showboxdepth=1|:
^^{ligature} ^^{em-dash}
\begintt
\hbox(4.30554+1.94444)x21.0, glue set 0.5
.\hbox(4.30554+1.94444)x5.0 []
.etc.
\endtt
And if you set |\showboxdepth=0|, you get only the top level:
\begintt
\hbox(4.30554+1.94444)x21.0, glue set 0.5 []
\endtt
(Notice how `^|[]|' and `^|etc.|'~indicate that the data has been
truncated.)

\danger A nonempty hbox is considered ``^{overfull}'' if its ^{glue}
cannot shrink to achieve the specified size, provided that ^|\hbadness| is
less than~100 or that the excess width (after shrinking by the maximum
amount) is more than ^|\hfuzz|. It is ``^{tight}'' if its glue shrinks and
the ^{badness} exceeds |\hbadness|; it is ``^{loose}'' if its glue
stretches and the badness exceeds |\hbadness| but is not greater than~100;
it is ``^{underfull}'' if its glue stretches and the badness is greater
than |\hbadness| and greater than~100. Similar remarks apply to nonempty
vboxes. \TeX\ prints a warning message and displays the offending box,
whenever such anomalies are discovered. Empty boxes are never considered
to be anomalous.

\ddanger When an ^{alignment} is ``overfull'' or ``tight'' or ``loose'' or
``underfull,'' you don't get a warning message for every aligned line;
you get only one message, and \TeX\ displays a {\sl^{prototype row}\/}
(or, with ^|\valign|, a {\sl prototype column\/}).
For example, suppose you say `|\tabskip=0pt plus10pt
\halign to200pt{&#\hfil\cr...\cr}|', ^^|\halign| and suppose that the aligned
material turns out to make two columns of widths $50\pt$ and $60\pt$,
respectively. Then you get the following message:
\begintt
Underfull \hbox (badness 2698) in alignment at lines 11--18
 [] []
\hbox(0.0+0.0)x200.0, glue set 3.0
.\glue(\tabskip) 0.0 plus 10.0
.\unsetbox(0.0+0.0)x50.0
.\glue(\tabskip) 0.0 plus 10.0
.\unsetbox(0.0+0.0)x60.0
.\glue(\tabskip) 0.0 plus 10.0
\endtt
The ``unset boxes'' in a prototype row show the individual column widths.
In this case the ^{tabskip glue} has to stretch 3.0 times its stretchability,
in order to reach the $200\pt$ goal, so the box is underfull. \ (According
to the formula in Chapter~14, the badness of this situation is~2700; \TeX\
actually uses a similar but more efficient formula, so it computes a
badness of~2698.) \ Every line of the alignment will be underfull, but
only the prototype row will be displayed in a warning message.
``^{Overfull rules}'' are never appended to the lines of overfull alignments.

\ddanger The |\tracing...|\ commands put all of their output into your log
file, unless the ^|\tracingonline| parameter is positive; in the latter
case, all diagnostic information goes to the terminal as well as to the
log file.  Plain \TeX\ has a ^|\tracingall| macro that turns on the
maximum amount of tracing of all kinds. It not only sets~up
|\tracingcommands|, |\tracingrestores|, |\tracingparagraphs|, and so on,
it also sets |\tracingonline=1|, and it sets ^|\showboxbreadth| and
^|\showboxdepth| to extremely high values, so that the entire contents of
all boxes will be displayed.

\ddanger Some production versions of \TeX\ have been streamlined for
speed. These implementations don't look at the values of the parameters
|\tracingparagraphs|, |\tracingpages|, |\tracingstats|, and
|\tracingrestores|, because \TeX\ runs faster when it doesn't have
to maintain statistics or keep tabs on whether tracing is required.
If you want all of \TeX's diagnostic tools, you should be sure to
use the right version.

\ddanger If you set ^|\pausing||=1|, \TeX\ will give you a chance to edit
each line of input as it is read from the file. In this way you can
make temporary patches (e.g., you can insert |\show...|\ commands)
while you're troubleshooting, without changing the actual contents
of the file, and you can keep \TeX\ running at human speed.

Final hint: When working on a long manuscript, it's best to prepare
only a few pages at a time. Set up a ``^{galley}'' file and a ``^{book}''
file, and enter your text on the galley file. \ (Put control
information that sets up your basic format at the beginning of this
file; an example of |galley.tex| appears in Appendix~E\null.) \
After the galleys come out looking right, you can append them to the
book file; then you can run the book file through \TeX\ occasionally,
in order to see how the pages really fit together. For example,
when the author prepared this manual, he did one chapter at a time,
and the longer chapters were split into subchapters.

\ddangerexercise Final exercise: Find all of the ^{lies} in this
manual, and all of the ^{jokes}.
\answer If this exercise isn't just a joke, the title of this
appendix is a lie.

\line{Final exhortation: G{\sc O} {\sc FORTH} now and create
{\sl masterpieces of the publishing art!\/}}

\endchapter

Who can understand his errors?
\author ^^{Biblical}{\sl Psalm 19\thinspace:\thinspace12\/} (c.~1000 B.C.)

\bigskip

It is one thing, to shew a Man that he is in an Error,
and another, to put him in possession of Truth.
\author JOHN ^{LOCKE}, {\sl An Essay Concerning Humane Understanding\/} (1690)
  % bk 4 ch 7 sec 11

\eject
\beginchapter Appendix A. Answers to\\All the\\Exercises

The preface to this manual points out the wisdom of trying to figure out
each exercise before you look up the answer here. But these answers are intended
to be read, since they occasionally provide additional information that
you are best equipped to understand when you have just worked on a problem.

\immediate\closeout\ans % this makes the answers file ready
\ninepoint
\input answers

\endchapter

If you can't solve a problem,
you can always look up the answer.
But please, try first to solve it by yourself;
then you'll learn more and you'll learn faster.
\author DONALD E. ^{KNUTH}, {\sl The \TeX book\/} (1984)

\bigskip

How answer you for your selues?
\author WILLIAM ^{SHAKESPEARE}, {\sl Much Adoe About Nothing\/} (1598)
  % Act IV, Scene 2, line 25

\eject
\beginchapter Appendix B. Basic\\Control\\Sequences

Let's begin this appendix with a chart that summarizes plain \TeX's
^^{summary of plain TeX} ^^|\+| ^^{tabbing} ^^{accents}
conventions.

\def\sep{\medskip\hrule width\hsize\medskip}
\medskip\smallskip
\hrule height .61803pt
\kern 1pt
\hrule
\medskip
\line{\strut Characters that are reserved for special purposes:\hfil
  |\|\hfil|{|\hfil|}|\hfil|$|\hfil|&|\hfil|#|\hfil|%|\hfil|^|\hfil|_|\hfil|~|}
\sep
\halign to\hsize{\strut\hfil#\hfil\tabskip\z@ plus10pt&
\hfil#\hfil&\hfil#\hfil&
\hfil#\hfil\tabskip\z@skip\cr
|\rm roman,|&|{\sl slanted},|&|{\bf boldface},|&|{\it italic\/} type|\cr
roman,&{\sl slanted},&{\bf boldface},&{\it italic\/} type\cr}
\sep
\halign to\hsize{\strut\hfil#\hfil\tabskip\z@ plus10pt&
\hfil#\hfil&\hfil#\hfil&\hfil#\hfil&\hfil#\hfil&\hfil#\hfil&
\hfil#\hfil&\hfil#\hfil&\hfil#\hfil&\hfil#\hfil&\hfil#\hfil&
\hfil#\hfil&\hfil#\hfil&\hfil#\hfil&\hfil#\hfil&\hfil#\hfil&
\hfil#\hfil&\hfil#\hfil&
\hfil#\hfil\tabskip\z@skip\cr
|``|&|''|&|--|&|---|&|?||`|&|!||`|&|\$|&|\#|&|\&|&|\%|&
  |\ae|&|\AE|&|\oe|&|\OE|&|\aa|&|\AA|&|\ss|&|\o|&|\O|\cr
``&''&--&---&?`&!`&\$&\#&\&&\%&\ae&\AE&\oe&\OE&\aa&\AA&\ss&\o&\O\cr}
\sep
\halign to\hsize{\strut\hfil#\hfil\tabskip\z@ plus10pt&
\hfil#\hfil&\hfil#\hfil&\hfil#\hfil&\hfil#\hfil&\hfil#\hfil&
\hfil#\hfil&\hfil#\hfil&\hfil#\hfil&\hfil#\hfil&\hfil#\hfil&
\hfil#\hfil&\hfil#\hfil&
\hfil#\hfil\tabskip\z@skip\cr
|\`a|&|\'e|&|\^o|&|\"u|&|\=y|&|\~n|&|\.p|&|\u\i|&
  |\v s|&|\H\j|&|\t\i u|&|\b k|&|\c c|&|\d h|\cr
\`a&\'e&\^o&\"u&\=y&\~n&\.p&\u\i&\v s&\H\j&\t\i u&\b k&\c c&\d h\cr}
\sep
\halign to\hsize{\strut\hfil#\hfil\tabskip\z@ plus10pt&
\hfil#\hfil&\hfil#\hfil&\hfil#\hfil&\hfil#\hfil&\hfil#\hfil&
\hfil#\hfil&\hfil#\hfil&\hfil#\hfil&\hfil#\hfil&
\hfil#\hfil\tabskip\z@skip\cr
|\l|&|\L|&|\dag|&|\ddag|&|\S|&|\P|&|{\it\$|&
  |\&}|&|\copyright|&|\TeX|&|\dots|\cr
\l&\L&\dag&\ddag&\S&\P&\phantom{\tt////}\it\$&
  \it\&\phantom{\tt/}&\copyright&\TeX&\dots\cr}
\sep
\line{\strut Line break controls:\hfil
  |\break|\hfil|\nobreak|\hfil|\allowbreak|\hfil|\hbox{unbreakable}|}
\line{\strut|dis\-cre\-tion\-ary hy\-phens|\hfil|virgule\slash breakpoint|}
\sep
\settabs 2\columns
\+\strut Breakable horizontal spaces:& Unbreakable horizontal spaces:\cr
\+|\|\] \ normal interword space& |~| \ normal interword space\cr
\+|\enskip| \ this\enskip much& |\enspace     | \ this\enspace much\cr
\+|\quad  | \ this\quad much&   |\thinspace   | \ this\thinspace much\cr
\+|\qquad | \ this\qquad much&  |\negthinspace| \ this\negthinspace much\cr
\+\strut|\hskip| \<arbitrary dimen>& |\kern| \<arbitrary dimen>\cr
\sep
\smallskip
\+\strut Vertical spaces:\hfill
  |\smallskip| $\vcenter{\hrule width2em\smallskip\hrule}$\hfill&
  |\medskip| $\vcenter{\hrule width3em\medskip\hrule}$\hfill
  |\bigskip| $\vcenter{\hrule width4em\bigskip\hrule}$&\cr
\smallskip
\sep
\line{\strut Page break controls:\hfil|\eject|\hfil|\supereject|\hfil
  |\nobreak|\hfil|\goodbreak|\hfil|\filbreak|}
\+\strut Vertical spaces and good breakpoints:&
  |\smallbreak|\hfill|\medbreak|\hfill|\bigbreak|&\cr
\sep
\settabs 4 \columns
\hbox{\strut|\settabs 4 \columns|}
\hbox{|\+Here's an example&of\hfill some &tabbing:&\hrulefill&\cr|}
\+Here's an example&of\hfill some &tabbing:&\hrulefill&\cr
\+\hfill|\hrulefill |&\hrulefill&\hfill|\dotfill |&\dotfill&\cr
\+\hfill|\leftarrowfill |&\leftarrowfill&
  \hfill|\rightarrowfill |&\rightarrowfill&\cr
\+\hfill|\upbracefill |&\upbracefill&
  \hfill|\downbracefill |&\downbracefill&\cr
\smallskip
\line{\strut More general alignments use |\halign|, |\valign|, |\omit|,
  |\span|, and |\multispan|.}
\sep
\line{\strut Examples of the principal conventions for text layout
  appear on the next page.}
\sep\unskip
\kern 1pt
\hrule height .61803pt
\eject
\begingroup\parindent\z@ \obeylines \baselineskip12pt plus 1pt
|% This test file generates the output shown on the opposite page.|
|% It's a bit complex because it tries to illustrate lots of stuff.|
|% TeX ignores commentary (like this) that follows a `%' sign.|
| |
|% First the standard output style is changed slightly:|
^|\hsize||=29pc % The lines in this book are 29 picas wide.|
^|\vsize||=42pc % The page body is 42 picas (not counting footlines).|
^|\footline||={\tenrm Footline\quad\dotfill\quad Page \folio}|
^|\pageno||=1009 % This is the starting page number (don't ask why).|
|% See Chapter 23 for the way to make other page format changes via|
|% \hoffset, \voffset, \nopagenumbers, \headline, or \raggedbottom.|
| |
^|\topglue|| 1in % This makes an inch of blank space (1in=2.54cm).|
^|\centerline||{\bf A Bold, Centered Title}|
^|\smallskip|| % This puts a little extra space after the title line.|
^|\rightline||{\it avec un sous-titre \`a la fran\c caise}|
|% Now we use \beginsection to introduce part 1 of the document.|
^|\beginsection|| 1. Plain \TeX nology % The next line must be blank!|
| |^^{blank line}
|The first paragraph of a new section is not indented.|
^|\TeX||\ recognizes the end of a paragraph when it comes to a blank|
|line in your manuscript file. % or to a `\par': See below.|^^|\par|
| |^^|\footnote|^^|\tt|^^|\char|
|Subsequent paragraphs {\it are\/} indented.\footnote*{The amount|
|    of indentation can be changed by changing a parameter called|
|{\tt\char`\\parindent}. Turn the page for a summary of \TeX's most|
|important parameters.} (See?) The computer breaks a paragraph's|
|text into lines in an interesting way---see reference~[1]---and h%|
|     yphenates words automatically     when necessary.|^^{percent}
| |
^|\midinsert|| % This begins inserted material, e.g., a figure.|
|\narrower\narrower % This brings the margins in (see Chapter 14).|
^|\noindent|| |^|\llap||{``}If there hadn't been room for this material on|
|the present page, it would have been inserted on the next one.''|
|\endinsert % This ends the insertion and the effect of \narrower.|
| |
^|\proclaim|| Theorem T. The typesetting of $math$ is discussed in|
|Chapters 16--19, and math symbols are summarized in Appendix~F.|
| |
|\beginsection 2. Bibliography\par % `\par' acts like a blank line.|
^|\frenchspacing|| % (Chapter 12 recommends this for bibliographies.)|
^|\item||{[1]} D.~E. Knuth and M.~F. Plass, ``Breaking paragraphs|^^{Plass}
|into lines,'' {\sl Softw. pract. exp. \bf11} (1981), 1119--1184.|^^{Knuth}
^|\bye|| % This is the way the file ends, not with a \bang but a \bye.|
\eject\endgroup
{\parindent 20pt
\topglue 1in % This makes an inch of blank space (1in=2.54cm).
\centerline{\bf A Bold, Centered Title}
\smallskip % This puts a little extra space between lines here.
\rightline{\it avec un sous-titre \`a la fran\c caise}
% Now we use \beginsection to introduce part 1 of the document.
\beginsection 1. Plain \TeX nology % The next line must be blank!

The first paragraph of a new section is not indented.
\TeX\ recognizes the end of a paragraph when it comes to a blank
line in your manuscript file. % or to a `\par': See below.

Subsequent paragraphs {\it are\/} indented.*{\spacefactor=3000}
(See?) The computer breaks a paragraph's
text into lines in an interesting way---see reference~[1]---and h%
     yphenates words automatically     when necessary.

\midinsert % This begins inserted material, e.g., a figure.
\narrower\narrower % This brings the margins in (see Chapter 14).
\noindent \llap{``}If there hadn't been room for this material on
the present page, it would have been inserted on the next one.''
\endinsert % This ends the insertion and the effect of \narrower.

\proclaim Theorem T. The typesetting of $math$ is discussed in
Chapters 16--19, and math symbols are summarized in Appendix~F.

\beginsection 2. Bibliography\par% `\par' is just like blank line.
\frenchspacing % (Chapter 12 recommends this for bibliographies.)
\item{[1]} D.~E. Knuth and M.~F. Plass, ``Breaking paragraphs
into lines,'' {\sl Softw. pract. exp. \bf11} (1981), 1119--1184.
\vfill
\hrule width 2in \kern 2.6pt
\textindent{*}\strut
The amount of indentation can be changed by changing a parameter called
{\tt\char`\\parindent}. Turn the page for a summary of \TeX's most
important parameters.\par
\baselineskip 24pt
\line{Footline\quad\dotfill\quad Page 1009}
\eject}

The preceding example illustrates most of the basic things
that you can do directly with plain \TeX, but it does not provide an
exhaustive list. Thus, it uses |\centerline| and |\rightline|,
but not |\leftline| or |\line|; it uses |\midinsert|, but not
|\topinsert| or |\pageinsert|; it uses |\smallskip|, but not
|\medskip| or |\bigskip|; it uses |\llap| but not |\rlap|, |\item|
but not |\itemitem|, |\topglue| but not |\hglue|. It does not
illustrate |\raggedright| setting of paragraphs; it does not use
^|\obeylines| or ^|\obeyspaces| to shut off \TeX's automatic
formatting. ^^{as is, see obeylines, obeyspaces}
All such control sequences are explained later in this appendix, and
further information can be found by looking them up in the index.
The main purpose of the example is to serve as a reminder of the
repertoire of possibilities.

Most of the control sequences used in the example are defined by
macros of plain \TeX\ format; but some of them are ^{primitive} (built in),
such as `|\par|' (end of paragraph), `|\noindent|' (beginning of
non-indented paragraph), and `|\/|' (italic correction). The example
also assigns values to two of \TeX's primitive parameters, namely
|\hsize| and |\vsize|. \TeX\ has scores of parameters, all of which
are listed in Chapter~24, but only a few of them are of special
concern to the majority of \TeX\ users. Here are examples of how
you might want to give new values to the most important ^{parameters}
other than |\hsize| and |\vsize|:
\par\begingroup\nobreak\medskip\parindent\z@ \obeylines
^|\tolerance||=500|\quad(\TeX\ will tolerate lines whose badness is %
  rated 500 or less.)
^|\looseness||=1|\quad(The next paragraph will be one line longer than %
  usual.)
^|\parindent||=4mm|\quad(Paragraphs will be indented by four millimeters.)
^|\hoffset||=1.5in|\quad(All output will be shifted right by one and a %
  half inches.)
^|\voffset||=24pt|\quad(All output will be shifted down by 24 points.)
^|\baselineskip||=11pt plus.1pt|\quad(Baselines will be $11\pt$ apart, %
  or a bit more.)
^|\parskip||=3pt plus1pt minus.5pt|\quad(Extra space will %
  precede each paragraph.)
\endgroup\medbreak\noindent
Plain \TeX\ uses |\parindent| also to control the amount of
indentation provided by ^|\item|, ^|\itemitem|, and ^|\narrower|.

\danger The remainder of this appendix is devoted to the details of the
plain \TeX\ format, which is a set of macros that come with normal
implementations of \TeX. These macros serve three basic purposes:
\ (1)~They make \TeX\ usable, because \TeX's primitive capabilities
operate at a very low level. A~``virgin'' \TeX\ system that has no
macros is like a newborn baby that has an immense amount to learn about
the real world; but it is capable of learning fast. \ (2)~The plain
\TeX\ macros provide a basis for more elaborate and powerful formats
tailored to individual tastes and applications. You can do a lot with
plain \TeX, but pretty soon you'll want to do even more. \ (3)~The macros
also serve to illustrate how additional formats can be designed.

\ninepoint
Somewhere in your computer system you should be able to find a file
called ^|plain.tex| that contains exactly what has been preloaded into
the running \TeX\ system that you use. Our purpose in the rest of this
appendix will be to discuss the contents of |plain.tex|. However, we
will not include a verbatim description, because some parts of that file
are too boring, and because the actual macros have been ``^{optimized}''
^^{efficiency}
with respect to memory space and running time. Unoptimized versions of the
macros are easier for humans to understand, so we shall deal with those;
|plain.tex| contains equivalent constructions that work better on a machine.

So here's the plan for the rest of Appendix~B\null: We shall go through
the contents of |plain.tex|, interspersing an edited transcription of
that file with comments about noteworthy details. When we come to
macros whose usage has not yet been explained---for example, somehow
|\vglue| and |\beginsection| never made it into Chapters 1 through~27---we
shall consider them from a user's viewpoint; but most of the time we shall
be addressing the issues from the standpoint of a macro designer.

\subsection The code tables. A format's first duty is to establish
^|\catcode| values. This is necessary because, for example, a |\def|
command can't be used until there are characters like |{| and~|}| of
categories 1 and~2. The ^|INITEX| program (which reads |plain.tex|
so that \TeX\ can be initialized) begins without knowing any grouping
characters; hence |plain.tex| starts out as follows:
\beginlines
|% This is the plain TeX format that's described in The TeXbook.|
|% N.B.: A version number is defined at the very end of this file;|
|%       please change that number whenever the file is modified!|
|% And don't modify the file under any circumstances.|
\smallbreak
|\catcode`\{=1 % left brace is begin-group character|
|\catcode`\}=2 % right brace is end-group character|
|\catcode`\$=3 % dollar sign is math shift|
|\catcode`\&=4 % ampersand is alignment tab|
|\catcode`\#=6 % hash mark is macro parameter character|
|\catcode`\^=7 \catcode`\^^K=7 % circumflex and uparrow for superscripts|
|\catcode`\_=8 \catcode`\^^A=8 % underline and downarrow for subscripts|
|\catcode`\^^I=10 % ASCII tab is treated as a blank space|
|\chardef\active=13 \catcode`\~=\active % tilde is active|
|\catcode`\^^L=\active \outer\def^^L{\par} % ASCII form-feed is \outer\par|
\smallskip
|\message{Preloading the plain format: codes,}|
\endlines
These instructions set up the nonstandard characters
|^^K| and |^^A| for ^{superscripts} and ^{subscripts}, in addition to~|^|
and~|_|, so that people with extended character sets can use
{\tentex\char'13} and {\tentex\char1} as recommended in Appendix~C\null.
^^{uparrow char} ^^{downarrow char}
Furthermore |^^I| (^{ASCII} ^\<tab>) is given category~10 (space);
and |^^L| (ASCII
\<formfeed>) becomes an active character that will detect runaways on
files that have been divided into ``^{file pages}'' by ^\<formfeed>
characters. The control sequence ^|\active| is defined to yield the
constant~13; this is the one category code that seems to deserve a
symbolic name, in view of its frequent use in constructing
special-purpose macros.

When |INITEX| begins, category 12 (other) has been assigned to all~256
possible characters, except that the 52~letters |A...Z| and |a...z| are
category~11 (letter), and a few other assignments equivalent to the
following have been made:
\begintt
\catcode `\\ =0   \catcode`\  =10   \catcode `\% =14
\catcode`\^^@=9   \catcode`\^^M=5   \catcode`\^^?=15
\endtt
Thus `|\|' is already an escape character, ^^{backslash}
`\]' is a space, ^^\<space>
and `|%|' is available for comments on the first line of the file;
ASCII ^\<null> is ignored, ASCII ^\<return> is an end-of-line character,
and ASCII ^\<delete> is invalid.

The ^|\message| command shown above prints a progress report on the terminal
when |plain.tex| is being input by |INITEX|. Later on comes
`|\message{registers,}|' and several other messages, but we won't mention
them specifically. The terminal will eventually display something like
this when initialization is complete:
\begintt
** plain
(plain.tex Preloading the plain format: codes, registers,
parameters, fonts, more fonts, macros, math definitions,
output routines, hyphenation (hyphen.tex))
* \dump
Beginning to dump on file plain.fmt
\endtt
followed by a variety of statistics about what fonts were loaded, etc.
If you want to make a new format |super.tex| that adds more features
to |plain.tex|, it's best not to make a new file containing all the
plain stuff, or even to |\input plain|; just type `|&plain super|' in
response to |INITEX|'s ^|**| prompt, ^^{ampersand} ^^|fmt|
to input |plain.fmt| at high speed.

After the opening |\message|, |plain.tex| goes on to define a control
sequence ^|\dospecials| that lists all the characters whose catcodes should
probably be changed to~12 (other) when copying things verbatim:
\beginlines
|\def\dospecials{\do\ \do\\\do\{\do\}\do\$\do\&%|
|  \do\#\do\^\do\^^K\do\_\do\^^A\do\%\do\~}|
\endlines
(Appendix E illustrates how to use |\dospecials|.) \ The ASCII codes for
\<null>, \<tab>, \<linefeed>, \<formfeed>, \<return>, and \<delete>
have not been included in the list.

At this point |plain.tex| completes its initialization of category codes
by setting |\catcode`\@=11|, thereby making the character `|@|' behave
temporarily like a letter. The command |\catcode`\@=12| will appear later,
hence ^{at-sign characters} will act just like ordinary punctuation
marks when \TeX\ is running. The idea is to make it easy for plain \TeX\
to have ^{private control sequences} that cannot be redefined by ordinary
users; all such control sequences will have at least one `|@|' in
their names.

The next job is to set up the ^|\mathcode| table:
\beginlines
|\mathcode`\^^@="2201 \mathcode`\^^A="3223 \mathcode`\^^B="010B|
|\mathcode`\^^C="010C \mathcode`\^^D="225E \mathcode`\^^E="023A|
|\mathcode`\^^F="3232 \mathcode`\^^G="0119 \mathcode`\^^H="0115|
|\mathcode`\^^I="010D \mathcode`\^^J="010E \mathcode`\^^K="3222|
|\mathcode`\^^L="2206 \mathcode`\^^M="2208 \mathcode`\^^N="0231|
|\mathcode`\^^O="0140 \mathcode`\^^P="321A \mathcode`\^^Q="321B|
|\mathcode`\^^R="225C \mathcode`\^^S="225B \mathcode`\^^T="0238|
|\mathcode`\^^U="0239 \mathcode`\^^V="220A \mathcode`\^^W="3224|
|\mathcode`\^^X="3220 \mathcode`\^^Y="3221 \mathcode`\^^Z="8000|
|\mathcode`\^^[="2205 \mathcode`\^^\="3214 \mathcode`\^^]="3215|
|\mathcode`\^^^="3211 \mathcode`\^^_="225F \mathcode`\^^?="1273|
|\mathcode`\ ="8000   \mathcode`\!="5021   \mathcode`\'="8000|
|\mathcode`\(="4028   \mathcode`\)="5029   \mathcode`\*="2203|
|\mathcode`\+="202B   \mathcode`\,="613B   \mathcode`\-="2200|
|\mathcode`\.="013A   \mathcode`\/="013D   \mathcode`\:="303A|
|\mathcode`\;="603B   \mathcode`\<="313C   \mathcode`\=="303D|
|\mathcode`\>="313E   \mathcode`\?="503F   \mathcode`\[="405B|
|\mathcode`\\="026E   \mathcode`\]="505D   \mathcode`\_="8000|
|\mathcode`\{="4266   \mathcode`\|\||="026A   \mathcode`\}="5267|
\endlines
A mathcode is relevant only when the corresponding category code is
11 or~12; therefore many of these codes will rarely be looked at. For
example, the math code for |^^M| specifies the character |\oplus|,
but it's hard to imagine a user who would want |^^M| (ASCII ^\<return>)
to produce an~$\oplus$ sign in the middle of a math formula, since plain
\TeX\ appends |^^M| to the end of every line of input. The math codes
have been set up here, however, to be entirely consistent with the
extended character set presented in Appendix~C and the Computer Modern
fonts described in Appendix~F\null. |INITEX| has done the rest of the
work, as far as mathcodes are concerned: It has set |\mathcode|$\,x=
x+\hex{7000}$ for each of the ten digits $x={}$|`0| to |`9|;
|\mathcode|$\,x=x+\hex{7100}$ for each of the 52~letters; and
|\mathcode|$\,x=x$ for all other values of~$x$.

There's no need to change the ^|\uccode| and ^|\lccode| tables. |INITEX|
has made |\uccode`X=`X|, |\uccode`x=`X|, |\lccode`X=`x|, |\lccode`x=`x|,
and it has made similar assignments for all other letters. The codes are
zero for all nonletters. These tables are used by \TeX's ^|\uppercase|
and |\lowercase| operations, and the hyphenation algorithm also looks
at |\lccode| (see Appendix~H\null). Changes should be made only in format
packages that set \TeX\ up for languages with more than~26 letters
(see Chapter~8).

Next comes the ^|\sfcode| table, which |INITEX| has initialized entirely
to 1000, except that |\sfcode`X=999| for each of the 26 uppercase letters.
Some characters are made ``transparent'' by setting
\beginlines
|\sfcode`\)=0 \sfcode`\'=0 \sfcode`\]=0 % won't change the space factor|
\endlines
and the |\nonfrenchspacing| macro will be used later to change the sfcodes
of special punctuation marks. \ (Chapter~12 explains what an |\sfcode| does.)

The last code table is called ^|\delcode|, and again it's necessary to
change only a few values. |INITEX| has made all delimiter codes equal to
$-1$, which means that no characters are recognized as delimiters in
formulas. But there's an exception: The value |\delcode`\.=0| has been
prespecified, so that `|.|'\ ^^{period} stands for a ``^{null delimiter}.''
\ (See Chapter~17.) \ Plain format sets up the following nine values,
based on the delimiters available in Computer Modern:
\beginlines
|\delcode`\(="028300    \delcode`\/="02F30E    \delcode`\)="029301|
|\delcode`\[="05B302    \delcode`\|\||="26A30C    \delcode`\]="05D303|
|\delcode`\<="26830A    \delcode`\\="26E30F    \delcode`\>="26930B|
\endlines
It's important to note that |\delcode`\{| and |\delcode`\}| have been left
equal to~$-1$.
^^{left brace} ^^{right brace}
^{Braces} are used for grouping, when supplying
arguments to macros; so they cannot also be used as math delimiters, or as
arguments to macros such as |\big|. (One could change their catcodes
to~12, and use some other pair of characters for grouping; but that
would not be plain \TeX.)

At this point the |plain.tex| file contains several definitions
\beginlines
|\chardef\@ne=1 \chardef\tw@=2 \chardef\thr@@=3 \chardef\sixt@@n=16|
|\chardef\@cclv=255 \mathchardef\@cclvi=256|
|\mathchardef\@m=1000 \mathchardef\@M=10000 \mathchardef\@MM=20000|
\endlines
which allow ``private'' control sequences ^|\@ne|, |\tw@|, etc.,
to be used as abbreviations for commonly used constants 1,~2,~\dots; this
convention makes \TeX\ run a little faster, ^^{optimization} ^^{efficiency}
and it means that the macros will consume slightly less memory space.
The usage of these abbreviations will not, however, be shown below
unless necessary; we shall pretend that `|1|\]' appears instead of\/ |\@ne|,
`|10000|\]' instead of\/ |\@M|, and so on, since that makes the programs
more readable. \ (Notice that the long form of\/ |\@ne| is `|1|\]' including
a space, because \TeX\ looks for and removes a space following a constant.)

\subsection Allocation of registers. The second major part of the
|plain.tex| file provides a foundation on which systems of independently
developed macros can coexist peacefully without interfering in their
usage of ^{registers}. The idea is that macro writers should abide by the
^^{macro conventions}
following ground rules: (1)~Registers numbered 0~to~9 are always free for
temporary ``^{scratch}'' use, but their values are always assumed to be
clobbered whenever any other macro might get into control.  \ (This applies
to registers like |\dimen0|, |\toks0|, |\skip1|, |\box3|, etc.; but \TeX\
has already reserved |\count0| through |\count9| for page number
identification.) \ (2)~The registers |\count255|, |\dimen255|, |\skip255|,
|\muskip255|, and |\toks255| are freely available in the same way.
\ (3)~All assignments to the scratch registers whose numbers are
1,~3, 5, 7, and~9 should be ^|\global|; all assignments to the
other scratch registers (0,~2, 4, 6, 8,~255) should be non-|\global|.
\ (This prevents the phenomenon of ``^{save stack buildup}'' discussed
in Chapter~27.)
\ (4)~Furthermore, it's possible to
use any register in a group, if you ensure that \TeX's grouping
mechanism will restore the register when you're done with the group, and
if you are certain that other macros will not make global assignments
to that register when you need it. \ (5)~But when a register is used
by several macros, or over long spans of time, it should be allocated
by |\newcount|, |\newdimen|, |\newbox|, etc. \ (6)~Similar remarks
apply to ^{input/output streams} used by ^|\read| and ^|\write|, to math
^{families} used by ^|\fam|, to sets of hyphenation rules used by
^|\language|, and to insertions (which require
^|\box|, ^|\count|, ^|\dimen|, and ^|\skip| registers all having the
same number).\looseness=-1

Some handy abbreviations are introduced at this point so that the macros
below will have easy access to scratch registers:
\beginlines
|\countdef\count@=255  \toksdef\toks@=0       \skipdef\skip@=0|
|\dimendef\dimen@=0    \dimendef\dimen@i=1    \dimendef\dimen@ii=2|
\endlines

Here now are the macros that provide allocation for quantities of more
permanent value. These macros use registers
|\count10| through |\count20| to hold the numbers that were allocated
most recently; for example, if\/ |\newdimen| has just reserved |\dimen15|,
the value of\/ |\count11| will be~15. However, the rest of the world
is not supposed to ``know'' that |\count11| has anything to do with
|\dimen| registers. There's a special counter called ^|\allocationnumber|
that will be equal to the most recently allocated number, after every
|\newcount|, |\newdimen|, \dots, |\newinsert| operation; macro packages
are supposed to refer to |\allocationnumber| if they want to find out what
number was allocated. The inside story of how allocation is actually performed
should be irrelevant when the allocation macros are used at a higher level;
you mustn't assume that |plain.tex| really does allocation in any
particular way.
\beginlines
|\count10=22 % this counter allocates \count registers 23, 24, 25, ...|
|\count11=9  % this counter allocates \dimen registers 10, 11, 12, ...|
|\count12=9  % this counter allocates \skip registers 10, 11, 12, ...|
|\count13=9  % this counter allocates \muskip registers 10, 11, 12, ...|
|\count14=9  % this counter allocates \box registers 10, 11, 12, ...|
|\count15=9  % this counter allocates \toks registers 10, 11, 12, ...|
|\count16=-1 % this counter allocates input streams 0, 1, 2, ...|
|\count17=-1 % this counter allocates output streams 0, 1, 2, ...|
|\count18=3  % this counter allocates math families 4, 5, 6, ...|
|\count19=0  % this counter allocates language codes 1, 2, 3, ...|
|\count20=255 % this counter allocates insertions 254, 253, 252, ...|
|\countdef\insc@unt=20 % nickname for the insertion counter|
|\countdef\allocationnumber=21 % the most recent allocation|
|\countdef|^|\m@ne||=22 \m@ne=-1 % a handy constant|
|\def|^|\wlog||{\immediate\write-1 } % this will write on log file (only)|
\smallbreak
|\outer\def|^|\newcount||{\alloc@0\count\countdef\insc@unt}|
|\outer\def|^|\newdimen||{\alloc@1\dimen\dimendef\insc@unt}|
|\outer\def|^|\newskip||{\alloc@2\skip\skipdef\insc@unt}|
|\outer\def|^|\newmuskip||{\alloc@3\muskip\muskipdef\@cclv}|
|\outer\def|^|\newbox||{\alloc@4\box\chardef\insc@unt}|
|\let\newtoks=\relax % this allows plain.tex to be read in twice|
|\outer\def\newhelp#1#2{\newtoks#1#1=\expandafter{\csname#2\endcsname}}|
|\outer\def|^|\newtoks||{\alloc@5\toks\toksdef\@cclv}|
|\outer\def|^|\newread||{\alloc@6\read\chardef\sixt@@n}|
|\outer\def|^|\newwrite||{\alloc@7\write\chardef\sixt@@n}|
|\outer\def|^|\newfam||{\alloc@8\fam\chardef\sixt@@n}|
|\outer\def|^|\newlanguage||{\alloc@9\language\chardef\@cclvi}|
\smallbreak
|\def\alloc@#1#2#3#4#5{\global\advance\count1#1 by 1|\parbreak%
|  \ch@ck#1#4#2% make sure there's still room|\parbreak%
|  \allocationnumber=\count1#1   \global#3#5=\allocationnumber|\parbreak%
|  \wlog{\string#5=\string#2\the\allocationnumber}}|
\smallbreak
|\outer\def|^|\newinsert||#1{\global\advance\insc@unt by-1|\parbreak%
|  \ch@ck0\insc@unt\count \ch@ck1\insc@unt\dimen|\parbreak%
|  \ch@ck2\insc@unt\skip  \ch@ck4\insc@unt\box|\parbreak%
|  \allocationnumber=\insc@unt|\parbreak%
|  \global\chardef#1=\allocationnumber|\parbreak%
|  \wlog{\string#1=\string\insert\the\allocationnumber}}|
\smallbreak
|\def\ch@ck#1#2#3{\ifnum\count1#1<#2%|\parbreak%
|  \else\errmessage{No room for a new #3}\fi}|^^|No room|
\endlines
The `|\alloc@|' macro does most of the work of allocation. It puts
a message like `|\maxdimen=\dimen10|' into the log file after
|\newdimen| has allocated a place for the |\dimen| register that
will be called |\maxdimen|; such information might be useful when
difficult macros are being ^{debugged}.

A ^|\newhelp| macro has been provided to aid in creating home-made help
texts: You can say, e.g., |\newhelp\helpout{This is a help message.}|, and
then give the command `^|\errhelp||=\helpout|' just before issuing an
^|\errmessage|. This method of creating help texts makes efficient
use of \TeX's ^{memory}, because it puts the text into a control sequence
name where it doesn't take up space that is needed for tokens.

The |plain| file now goes ahead and allocates registers for important constants:
\beginlines
|\newdimen\maxdimen \maxdimen=16383.99999pt|
|\newskip\hideskip \hideskip=-1000pt plus1fill|
|\newskip\centering \centering=0pt plus 1000pt minus 1000pt|
|\newdimen\p@ \p@=1pt % this saves macro space and time|
|\newdimen\z@ \z@=0pt| % likewise
|\newskip\z@skip \z@skip=0pt plus0pt minus0pt|
|\newbox\voidb@x % permanently void box register|
\endlines
The control sequence ^|\maxdimen| stands for the largest permissible
\<dimen>. Alignment macros that appear below will make use of special glue
values called ^|\hideskip| and ^|\centering|.  {\sl N.B.: These three constants
must not be changed under any circumstances\/}; you should either ignore
them completely or just use them and enjoy them.
In fact, the next four constant registers (^|\p@|, ^|\z@|, ^|\z@skip|,
and ^|\voidb@x|) have been given private names so that they are untouchable.
The control sequence |\p@| is used several dozen times as an abbreviation
for `|pt |', and |\z@| is used quite often to stand for either `|0pt |' or
`|0 |'; the use of such abbreviations saves almost 10\% of the space
needed to store the tokens in plain \TeX's macros.  But we shall stick to
^^{optimization}
the unabbreviated forms below, so that the macros are more readable.

A different sort of allocation comes next:
^^|\string|^^|\escapechar|^^|\uppercase|^^|\iftrue|^^|\iffalse|
\beginlines
|\outer\def\newif#1{\count@=\escapechar \escapechar=-1|\parbreak%
|  |^|\expandafter||\expandafter\expandafter|\parbreak%
|   \def\@if#1{true}{\let#1=\iftrue}%|\parbreak%
|  \expandafter\expandafter\expandafter|\parbreak%
|   \def\@if#1{false}{\let#1=\iffalse}%|\parbreak%
|  \@if#1{false}\escapechar=\count@} % the condition starts out false|\parbreak%
|\def\@if#1#2{|^|\csname||\expandafter\if@\string#1#2\endcsname}|\parbreak%
|{\uccode`1=`i \uccode`2=`f \uppercase{\gdef\if@12{}}} % `if' is required|
\endlines
For example, the command |\newif\ifalpha| creates a trio of control
sequences called |\alphatrue|, |\alphafalse|, and |\ifalpha|
(see Chapter~20).

\subsection Parameters. |INITEX| sets almost all of the numeric registers
and parameters equal to zero; it makes all of the token registers and parameters
empty; and it makes all of the box registers void. But there are a few
^^{parameters, default values} ^^{default values of parameters}
exceptions: ^|\mag| is set initially to~1000, ^|\tolerance| to~10000,
^|\maxdeadcycles| to~25, ^|\hangafter| to~1, ^|\escapechar| to~|`\\|, and
^|\endlinechar| to~|`\^^M|.  Plain \TeX\ assigns new parameter values as
follows:
\beginlines
^|\pretolerance||=100 |^|\tolerance||=200  |^|\hbadness||=1000 |%
  ^|\vbadness||=1000|
^|\linepenalty||=10 |^|\hyphenpenalty||=50 |^|\exhyphenpenalty||=50|
^|\binoppenalty||=700 |^|\relpenalty||=500|
^|\clubpenalty||=150 |^|\widowpenalty||=150 |^|\displaywidowpenalty||=50|
^|\brokenpenalty||=100 |^|\predisplaypenalty||=10000|
^|\doublehyphendemerits||=10000 |^|\finalhyphendemerits||=5000 |%
  ^|\adjdemerits||=10000|
^|\tracinglostchars||=1 |^|\uchyph||=1 |^|\delimiterfactor||=901|
^|\defaulthyphenchar||=`\- |^|\defaultskewchar||=-1 |%
  ^|\newlinechar||=-1|
^|\showboxbreadth||=5 |^|\showboxdepth||=3 |^|\errorcontextlines||=5|
\smallbreak
^|\hfuzz||=0.1pt |^|\vfuzz||=0.1pt |^|\overfullrule||=5pt|
^|\hsize||=6.5in |^|\vsize||=8.9in |^|\parindent||=20pt|
^|\maxdepth||=4pt |^|\splitmaxdepth||=\maxdimen |^|\boxmaxdepth||=\maxdimen|
^|\delimitershortfall||=5pt |^|\nulldelimiterspace||=1.2pt |%
  ^|\scriptspace||=0.5pt|
\smallbreak
^|\parskip||=0pt plus 1pt|
^|\abovedisplayskip||=12pt plus 3pt minus 9pt|
^|\abovedisplayshortskip||=0pt plus 3pt|
^|\belowdisplayskip||=12pt plus 3pt minus 9pt|
^|\belowdisplayshortskip||=7pt plus 3pt minus 4pt|
^|\topskip||=10pt |^|\splittopskip||=10pt|
^|\parfillskip||=0pt plus 1fil|
\smallbreak
^|\thinmuskip||=3mu|
^|\medmuskip||=4mu plus 2mu minus 4mu|
^|\thickmuskip||=5mu plus 5mu|
\endlines

(Some parameters are set by \TeX\ itself as it runs, so it is inappropriate
to initialize them: ^|\time|, ^|\day|, ^|\month|, and ^|\year| are established
at the beginning of a job; ^|\outputpenalty| is given a value when an
|\output| routine is invoked; ^|\predisplaysize|, ^|\displaywidth|,
and ^|\displayindent| get values just before a display is processed;
and the values ^|\looseness||=0|, ^|\hangindent||=0pt|, ^|\hangafter||=1|,
^|\parshape||=0| are assigned at the end of a paragraph and when \TeX\ enters
internal vertical mode.)

The parameters ^|\baselineskip|, ^|\lineskip|, and ^|\lineskiplimit| have not
been initialized here, but a macro called ^|\normalbaselines| is defined below;
this macro sets |\baselineskip=\normalbaselineskip|,
|\lineskip=\normallineskip|, and |\lineskiplimit=\normallineskiplimit|.
An indirect approach like this has been used so that several different
type sizes may be handled, as illustrated in Appendix~E\null.  Plain \TeX\
deals exclusively with $10\pt$ type, but it supports extension to other styles.

Some ``^{pseudo parameters}'' come next. These quantities behave just like
internal parameters of \TeX, and users are free to change them in the same
way, but they are part of the plain \TeX\ format
rather than primitives of the language. ^^{parameters, pseudo}
\beginlines
|\newskip|^|\smallskipamount|| % the amount of a \smallskip|
|  \smallskipamount=3pt plus1pt minus1pt|
|\newskip|^|\medskipamount|| % the amount of a \medskip|
|  \medskipamount=6pt plus2pt minus2pt|
|\newskip|^|\bigskipamount|| % the amount of a \bigskip|
|  \bigskipamount=12pt plus4pt minus4pt|
|\newskip|^|\normalbaselineskip|| % normal value of \baselineskip|
|  \normalbaselineskip=12pt|
|\newskip|^|\normallineskip|| % normal value of \lineskip|
|  \normallineskip=1pt|
|\newdimen|^|\normallineskiplimit|| % normal value of \lineskiplimit|
|  \normallineskiplimit=0pt|
|\newdimen|^|\jot|| % unit of measure for opening up displays|
|  \jot=3pt|
|\newcount|^|\interdisplaylinepenalty|| % interline penalty in \displaylines|
|  \interdisplaylinepenalty=100|
|\newcount|^|\interfootnotelinepenalty|| % interline penalty in footnotes|
|  \interfootnotelinepenalty=100|
\endlines

\subsection Font information. Now |plain.tex| brings in the data that \TeX\
needs to know about how to typeset lots of characters in lots of different
fonts. First the ^|\magstep| macros are defined, to support font scaling:
\beginlines
|\def|^|\magstephalf||{1095 }|
|\def\magstep#1{\ifcase#1 1000\or|
|  1200\or 1440\or 1728\or 2074\or 2488\fi\relax}|
\weakendlines
(Incidentally, |\magstep| doesn't use ^|\multiply| to compute values,
since it is supposed to expand to a ^\<number> en route to \TeX's
``stomach''; ^^{anatomy} |\multiply| wouldn't work, because
it is an assignment command, performed only in the stomach.)

One of the main things that distinguishes one format from another
is the fact that each format gives \TeX\ the necessary knowledge about
a certain family of typefaces. In this case the ^{Computer Modern} types
described in Appendix~F are taken as a basis, although there is a provision
for incorporating other styles.
\beginlines
|\font|^|\tenrm||=cmr10        \font\preloaded=cmr9     \font\preloaded=cmr8|
\nobreak
|\font\sevenrm=cmr7       \font\preloaded=cmr6     \font\fiverm=cmr5|
\smallskip
|\font\teni=cmmi10        \font\preloaded=cmmi9    \font\preloaded=cmmi8|
\nobreak
|\font\seveni=cmmi7       \font\preloaded=cmmi6    \font\fivei=cmmi5|
\smallbreak
|\font\tensy=cmsy10       \font\preloaded=cmsy9    \font\preloaded=cmsy8|
\nobreak
|\font\sevensy=cmsy7      \font\preloaded=cmsy6    \font\fivesy=cmsy5|
\nobreak\smallskip
|\font\tenex=cmex10|
\smallbreak
|\font\tenbf=cmbx10       \font\preloaded=cmbx9    \font\preloaded=cmbx8|
\nobreak
|\font\sevenbf=cmbx7      \font\preloaded=cmbx6    \font\fivebf=cmbx5|
\smallskip
|\font\tensl=cmsl10       \font\preloaded=cmsl9    \font\preloaded=cmsl8|
\nobreak
|\font\tentt=cmtt10       \font\preloaded=cmtt9    \font\preloaded=cmtt8|
\nobreak
|\font\tenit=cmti10       \font\preloaded=cmti9    \font\preloaded=cmti8|
\nobreak
|\font\preloaded=cmss10   \font\preloaded=cmssq8|
\nobreak
|\font\preloaded=cmssi10  \font\preloaded=cmssqi8|
\smallskip
|\font\preloaded=cmr7 scaled \magstep4 % for titles|
\nobreak
|\font\preloaded=cmtt10 scaled \magstep2|
\nobreak
|\font\preloaded=cmssbx10 scaled \magstep2|
\smallskip
|% Additional \preloaded fonts can be specified here.|
|% (And those that were \preloaded above can be eliminated.)|
|\let\preloaded=\undefined % preloaded fonts must be declared anew later.|
\endlines
Notice that most of the fonts have been called ^|\preloaded|; but the
control sequence |\preloaded| is made undefined at the very end, so those
fonts cannot be used directly. There are two reasons for this strange
approach: First, it is desirable to keep the total number of fonts
of plain \TeX\ relatively small, because plain \TeX\ is a sort of
standard format; it shouldn't cost much for people to acquire all the
fonts of plain \TeX\ in addition to the ones that they really want. Second, it
is desirable on many computer systems to preload the information for
most of the fonts that people will actually be using, since this saves
a lot of machine time. The |\preloaded| font
information goes into \TeX's memory, where it will come alive instantly
if the user defines the corresponding |\font| again. For example, the
book format in Appendix~E says `|\font\ninerm=cmr9|'; after that
assignment has been obeyed, the control sequence |\ninerm| will
identify the |cmr9| font, whose information does not have to be loaded again.

The exact number and nature of fonts that are preloaded is unimportant;
the only important thing needed for standardization between machines
is that sixteen basic fonts (|cmr10|, |cmr7|, \dots,~|cmti10|) should
actually be loaded. The |plain.tex| files used on different machines
can be expected to differ widely with respect to preloaded fonts,
since the choice of how many fonts to preload and the selection of the
most important fonts depends on local conditions.
For example, at the author's university it is useful to preload a
font that contains the Stanford seal, but that particular font is not
very popular at Berkeley.

Most of these fonts have the default values of\/ ^|\hyphenchar| and
^|\skewchar|, namely |`-| and |-1|; but the math italic and math symbol fonts
have special |\skewchar| values, which are defined next:
\beginlines
|\skewchar\teni='177 \skewchar\seveni='177 \skewchar\fivei='177|
|\skewchar\tensy='60 \skewchar\sevensy='60 \skewchar\fivesy='60|
\endlines

Once the fonts are loaded, they are also grouped into families for
use in math setting, and shorthand names like ^|\rm| and~^|\it| are defined:
\beginlines
|\textfont0=\tenrm \scriptfont0=\sevenrm |^|\scriptscriptfont||0=\fiverm|
\nobreak
|\def\rm{\fam0 \tenrm}|
^|\textfont||1=\teni  |^|\scriptfont||1=\seveni  \scriptscriptfont1=\fivei|
\nobreak
|\def|^|\mit||{\fam1 } \def|^|\oldstyle||{\fam1 \teni}|
|\textfont2=\tensy \scriptfont2=\sevensy \scriptscriptfont2=\fivesy|
\nobreak
|\def|^|\cal||{\fam2 }|
|\textfont3=\tenex \scriptfont3=\tenex   \scriptscriptfont3=\tenex|
\smallbreak
|\newfam\itfam \def\it{\fam\itfam\tenit} \textfont\itfam=\tenit|
|\newfam\slfam \def|^|\sl||{\fam\slfam\tensl} \textfont\slfam=\tensl|
|\newfam\bffam \def|^|\bf||{\fam\bffam\tenbf} \textfont\bffam=\tenbf|
\nobreak
|  \scriptfont\bffam=\sevenbf   \scriptscriptfont\bffam=\fivebf|
^|\newfam||\ttfam \def|^|\tt||{\fam\ttfam\tentt} \textfont\ttfam=\tentt|
\endlines

\subsection Macros for text. The fifth part of |plain.tex| introduces
macros that do basic formatting unrelated to mathematics. First come
some macros that were promised above:
\beginlines
|\def|^|\frenchspacing||{\sfcode`\.=1000 \sfcode`\?=1000 \sfcode`\!=1000|
\nobreak
|  \sfcode`\:=1000 \sfcode`\;=1000 |^|\sfcode||`\,=1000 }|
|\def|^|\nonfrenchspacing||{\sfcode`\.=3000 \sfcode`\?=3000 \sfcode`\!=3000|
\nobreak
|  \sfcode`\:=2000 \sfcode`\;=1500 \sfcode`\,=1250 }|
|\def|^|\normalbaselines||{\lineskip=\normallineskip|
\nobreak
|  \baselineskip=\normalbaselineskip \lineskiplimit=\normallineskiplimit}|
\endlines

The next macros are simple but vital. ^^{control tab}^^{control return}
^^{control space} First |\|\<tab> and |\|\<return> are defined so that they
expand to |\|\<space>; this helps to prevent confusion, since all three
cases look identical when displayed on most computer terminals.
Then the macros ^|\lq|, ^|\rq|, ^|\lbrack|, and ^|\rbrack| are defined,
for people who have difficulty typing quotation marks and/or brackets. The
control sequences ^|\endgraf|
and ^|\endline| are made equivalent to \TeX's primitive ^|\par| and~^|\cr|
operations, since it is often useful to redefine the meanings of\/ |\par|
and~|\cr| themselves. Then come the definitions of\/ ^|\space| (a blank space),
^|\empty| (a~list of no tokens), and ^|\null| (an~empty hbox). Finally,
^|\bgroup| and ^|\egroup| are made to provide ``implicit'' grouping characters
that turn out to be especially useful in macro definitions. \ (See Chapters
24--26 and Appendix~D for information about ^{implicit characters}.)
\beginlines
|\def\^^I{\ } \def\^^M{\ }|
|\def\lq{`} \def\rq{'} \def\lbrack{[} \def\rbrack{]}|
|\let\endgraf=\par \let\endline=\cr|
|\def\space{ } \def\empty{} \def\null{\hbox{}}|
|\let\bgroup={ \let\egroup=}|
\endlines

Something a bit tricky comes up now in the definitions of\/ ^|\obeyspaces|
and ^|\obeylines|, since \TeX\ is only ``half obedient'' while these
definitions are half finished:
\beginlines
|\def\obeyspaces{\catcode`\ =\active}|
|{\obeyspaces\global\let =\space}|
|{\catcode`\^^M=\active % these lines must end with `%'|
\nobreak
|  \gdef\obeylines{\catcode`\^^M=\active \let^^M=\par}%|
\nobreak
|  \global\let^^M=\par} % this is in case ^^M appears in a \write|
\endlines
The |\obeylines| macro says `|\let^^M=\par|' instead of `|\def^^M{\par}|'
because the |\let| technique allows constructions such as
`|\let\par=\cr| |\obeylines| ^|\halign||{...}|' in which ^|\cr|'s need
not be given within the alignment.

The ^|\loop||...|^|\repeat| macro provides for iterative operations
as illustrated at the end of Chapter~20. In this macro and several others,
the control sequence `^|\next|' is given a temporary value that is not
going to be needed later; thus, |\next| acts like a ``^{scratch control
sequence}.''
\beginlines
|\def\loop#1\repeat{\def\body{#1}\iterate}|
|\def\iterate{\body \let\next=\iterate \else\let\next=\relax\fi \next}|
|\let\repeat=\fi % this makes \loop...\if...\repeat skippable|
\endlines

Spacing is the next concern. The macros ^|\enskip|, ^|\quad|, and ^|\qquad|
provide spaces that are legitimate breakpoints within a paragraph;
^|\enspace|, ^|\thinspace|, and ^|\negthinspace| produce space that cannot
cause a break (although the space will disappear if it occurs just
next to certain kinds of breaks). All six of these spaces are relative
to the current font. You can get horizontal space that
never disappears by saying `^|\hglue|\<glue>'; this space is able to
stretch or shrink. Similarly, there's a vertical analog,
`^|\vglue|\<glue>'.
The ^|\nointerlineskip| macro suppresses interline glue that would
ordinarily be inserted before the next box in vertical mode; this is a
``one shot'' macro, but ^|\offinterlineskip| is more drastic---it sets
things up so that future ^{interline glue} will be present, but zero.
There also are macros for potentially breakable vertical spaces:
^|\smallskip|, ^|\medskip|, and ^|\bigskip|.^^|\topglue|
\beginlines
|\def\enskip{\hskip.5em\relax}    \def\enspace{\kern.5em }|
|\def\quad{\hskip1em\relax}       \def\qquad{\hskip2em\relax}|
|\def\thinspace{\kern .16667em }  \def\negthinspace{\kern-.16667em }|
\smallbreak
|\def\hglue{|^|\afterassignment||\hgl@\skip@=}|
|\def\hgl@{\leavevmode \count@=\spacefactor \vrule width0pt|
|  \nobreak\hskip\skip@ \spacefactor=\count@}|
|\def\vglue{\afterassignment\vgl@\skip@=}|
|\def\vgl@{\par \dimen@=\prevdepth \hrule height0pt|
|  \nobreak\vskip\skip@ \prevdepth=\dimen@}|
|\def\topglue{\nointerlineskip \vglue-\topskip \vglue} % for top of page|
\smallbreak
|\def\nointerlineskip{\prevdepth=-1000pt }|
|\def\offinterlineskip{\baselineskip=-1000pt|
|  \lineskip=0pt \lineskiplimit=\maxdimen}|
\smallbreak
|\def\smallskip{\vskip\smallskipamount}|
|\def\medskip{\vskip\medskipamount}|
|\def\bigskip{\vskip\bigskipamount}|
\endlines

Speaking of breakpoints, the following macros introduce ^{penalty} markers
that make breaking less, or more, desirable. The ^|\break|, ^|\nobreak|,
and ^|\allowbreak| macros are intended for use in any mode; the |~|~(tie)
and ^|\slash| (hyphen-like~`/') macros are intended for horizontal mode.
The others are intended only for vertical mode, i.e., between paragraphs,
so they begin with |\par|.
\beginlines
|\def\break{\penalty-10000 } \def\nobreak{\penalty10000 }|
|\def\allowbreak{\penalty0 }|
|\def~{\penalty10000\ }|
|\def\slash{/\penalty\exhyphenpenalty}|
\smallbreak
|\def|^|\filbreak||{\par\vfil\penalty-200\vfilneg}|
|\def|^|\goodbreak||{\par\penalty-500 }|
|\def|^|\eject||{\par\penalty-10000 }|
|\def|^|\supereject||{\par\penalty-20000 }|
\smallbreak
|\def|^|\removelastskip||{\ifdim\lastskip=0pt \else\vskip-\lastskip\fi}|
|\def|^|\smallbreak||{\par \ifdim\lastskip<\smallskipamount|\parbreak%
|  \removelastskip \penalty-50 \smallskip \fi}|
|\def|^|\medbreak||{\par \ifdim\lastskip<\medskipamount|\parbreak%
|  \removelastskip \penalty-100 \medskip \fi}|
|\def|^|\bigbreak||{\par \ifdim\lastskip<\bigskipamount|\parbreak%
|  \removelastskip \penalty-200 \bigskip \fi}|
\endlines

Boxes are next: ^|\line|, ^|\leftline|, ^|\rightline|, and ^|\centerline|
produce boxes of the full line width, while ^|\llap| and ^|\rlap| make
boxes whose effective width is zero. The ^|\underbar| macro puts its
argument into an hbox with a straight line at a fixed distance under it.
\beginlines
|\def\line{\hbox to\hsize}|
|\def\leftline#1{\line{#1\hss}} \def\rightline#1{\line{\hss#1}}|
|\def\centerline#1{\line{\hss#1\hss}}|
\smallskip
|\def\llap#1{\hbox to 0pt{\hss#1}} \def\rlap#1{\hbox to 0pt{#1\hss}}|
\smallskip
|\def\m@th{\mathsurround=0pt }|
|\def\underbar#1{$\setbox0=\hbox{#1} \dp0=0pt|\parbreak%
|  \m@th \underline{\box0}$}|
\endlines
(Notice that |\underbar| uses math mode to do its job, although the
operation is essentially non-mathematical in nature. A few of the other
macros below use math mode in similar ways; thus, \TeX's mathematical
abilities prove to be useful even when no mathematical typesetting
is actually being done. A special control sequence ^|\m@th| is
used to ``turn off'' ^|\mathsurround| when such constructions are
being performed.)

\smallbreak
A ^|\strut| is implemented here as a rule of width zero, since this
takes minimum space and time in applications where numerous struts are
present.
\beginlines
|\newbox\strutbox|
|\setbox\strutbox=\hbox{\vrule height8.5pt depth3.5pt width0pt}|
|\def\strut{\relax\ifmmode\copy\strutbox\else\unhcopy\strutbox\fi}|
\weakendlines
The `^|\relax|' in this macro and in others below is necessary in case
|\strut| appears first in an alignment entry, because \TeX\ is in a
somewhat unpredictable mode at such times (see Chapter~22).

The ^|\ialign| macro provides for ^{alignments} when it is necessary to
be sure that ^|\tabskip| is initially zero. The ^|\hidewidth| macro can be
used essentially as |\hfill| in alignment entries that are permitted to
``stick out'' of their column. There's also ^|\multispan|, which permits
alignment entries to span one or more columns.
\beginlines
|\def\ialign{\everycr={}\tabskip=0pt \halign} % initialized \halign|
|\def\hidewidth{\hskip\hideskip}|
\smallbreak
|\newcount\mscount|
|\def\multispan#1{\omit\mscount=#1\relax\loop\ifnum\mscount>1 \sp@n\repeat}|%
\kern-10pt\null
|\def\sp@n{\span\omit \advance\mscount by -1 }|
\endlines

Now we get to the ``^{tabbing}'' macros, which are more complicated than
anything else in plain \TeX. They keep track of the tab positions by
maintaining boxes full of empty boxes having the specified widths.
\ (The best way to understand these macros is probably to watch
them in action on simple examples, using |\tracingall|.)
\beginlines
|\newif\ifus@ \newif\if@cr|
|\newbox\tabs \newbox\tabsyet \newbox\tabsdone|
\smallbreak
|\def|^|\cleartabs||{\global\setbox\tabsyet=\null \setbox\tabs=\null}|
|\def|^|\settabs||{\setbox\tabs=\null \futurelet\next\sett@b}|
|\let\+=\relax % in case this file is being read in twice|
|\def\sett@b{\ifx\next\+\def\nxt{\afterassignment\s@tt@b\let\nxt}%|\parbreak%
|  \else\let\nxt=\s@tcols\fi|\parbreak%
|  \let\next=\relax \nxt} % turn off \outerness|
|\def\s@tt@b{\let\nxt=\relax \us@false\m@ketabbox}|
|\outer\def|^|\+||{\tabalign} \def\tabalign{\us@true \m@ketabbox}|
|\def\s@tcols#1\columns{\count@=#1 \dimen@=\hsize|\parbreak%
|  \loop \ifnum\count@>0 \@nother \repeat}|
|\def\@nother{\dimen@ii=\dimen@ \divide\dimen@ii by\count@|\parbreak%
|  \setbox\tabs=\hbox{\hbox to\dimen@ii{}\unhbox\tabs}%|\parbreak%
|  \advance\dimen@ by-\dimen@ii \advance\count@ by -1 }|
\smallbreak
|\def\m@ketabbox{\begingroup|\parbreak%
|  \global\setbox\tabsyet=\copy\tabs \global\setbox\tabsdone=\null|\parbreak%
|  \def\cr{\@crtrue\crcr\egroup\egroup|\parbreak%
|    \ifus@ \unvbox0 |^|\lastbox||\fi \endgroup|\parbreak%
|    \setbox\tabs=\hbox{\unhbox\tabsyet|^|\unhbox||\tabsdone}}%|\parbreak%
|  \setbox0=\vbox\bgroup\@crfalse \ialign\bgroup&\t@bbox##\t@bb@x\crcr}|
\smallbreak
|\def\t@bbox{\setbox0=\hbox\bgroup}|
|\def\t@bb@x{\if@cr\egroup % now \box0 holds the column|\parbreak%
|  \else\hss\egroup \global\setbox\tabsyet=\hbox{\unhbox\tabsyet|\parbreak%
|      \global\setbox1=\lastbox}% now \box1 holds its size|\parbreak%
|    \ifvoid1 \global\setbox1=\hbox to\wd0{}%|\parbreak%
|    \else\setbox0=\hbox to\wd1{\unhbox0}\fi|\parbreak%
|    \global\setbox\tabsdone=\hbox{\box1\unhbox\tabsdone}\fi|\parbreak%
|  \box0}|
\endlines
The macro |\+| has been declared `^|\outer|' here, so that \TeX\ will
be better able to detect runaway arguments and definitions (see Chapter~20).
A non-|\outer| version, called ^|\tabalign|, has also been provided
in case it is necessary to use |\+| in some ``inner'' place.
You can use |\tabalign| just like |\+|, except after ^|\settabs|.

\textindent{$\bullet$} Paragraph shapes of a limited but important kind
are provided by ^|\item|, ^|\itemitem|, and ^|\narrower|. There are also
two macros that haven't been mentioned before: \ (1)~^|\hang| causes
hanging indentation by the normal amount of\/ ^|\parindent|, after the
first line; thus, the entire paragraph will be indented by the same
amount (unless it began with ^|\noindent|). \ (2)~^|\textindent||{stuff}|
is like ^|\indent|, but it puts the `|stuff|' into the indentation, flush
right except for an en~space; it also removes spaces that might follow
the right brace in `|{stuff}|'. For example, the present paragraph
was typeset by the commands `|\textindent{$\bullet$} Paragraph shapes
...|'; the opening `P' occurs at the normal position for a paragraph's
first letter. ^^|\bullet| ^^|\ignorespaces|
\beginlines
|\def\hang{\hangindent\parindent}|
|\def\item{\par\hang\textindent}|
|\def\itemitem{\par\indent \hangindent2\parindent \textindent}|
|\def\textindent#1{\indent\llap{#1\enspace}\ignorespaces}|
|\def\narrower{\advance\leftskip by\parindent|\parbreak%
|  \advance\rightskip by\parindent}|
\endlines

The ^|\beginsection| macro is intended to mark the beginning of a
new major subdivision in a document; to use it, you say
`|\beginsection|\<section title>' followed by a blank line (or~|\par|).
The macro first emits glue and penalties, designed to start a new page
if the present page is nearly full; then it makes a ^|\bigskip| and
puts the section title flush left on a line by itself, in boldface type.
The section title is also displayed on the terminal.
After a ^|\smallskip|, with page break prohibited, a ^|\noindent|
command is given; this suppresses indentation in the next paragraph,
i.e., in the first paragraph of the new section.
\ (However, the next ``paragraph'' will be empty if vertical mode
material immediately follows the |\beginsection| command.) ^^|\message|
\beginlines
|\outer\def\beginsection#1\par{\vskip0pt plus.3\vsize\penalty-250|\parbreak%
|  \vskip0pt plus-.3\vsize\bigskip\vskip\parskip|\parbreak%
|  \message{#1}\leftline{\bf#1}\nobreak\smallskip\noindent}|
\endlines

Special statements in a mathematical paper are often called ^{theorems},
lemmas, definitions, axioms, postulates, remarks, corollaries, algorithms,
facts, conjectures, or some such things, and they generally are
given special typographic treatment. The ^|\proclaim| macro, which
was illustrated earlier in this appendix and also in Chapter~20,
puts the title of the proclamation in boldface, then sets the rest
of the paragraph in slanted type. The paragraph is followed by
something similar to ^|\medbreak|, except that the amount of penalty
is different so that page breaks are discouraged:
\beginlines
|\outer\def\proclaim #1. #2\par{\medbreak|\parbreak%
|  \noindent{\bf#1.\enspace}{\sl#2\par}%|\parbreak%
|  \ifdim\lastskip<\medskipamount \removelastskip\penalty55\medskip\fi}|
\endlines

^{Ragged-right setting} is initiated by restricting the spaces between
words to have a fixed width, and by putting variable space at the right
of each line. You should not call ^|\raggedright| until your text font
has already been specified; it is assumed that the ragged-right material
will not be in a variety of different sizes. \ (If this assumption is
not valid, a different approach should be used: ^|\fontdimen| parameters
3 and~4 of the fonts you will be using should be set to zero, by saying,
e.g., `|\fontdimen3\tenrm=0pt|'. These parameters specify the stretchability
and shrinkability of ^{interword spaces}.) \ ^^|\spaceskip| ^^|xspaceskip|
A special macro ^|\ttraggedright| should be used for ragged-right setting
in ^{typewriter type}, since the spaces between words are generally
bigger in that style. \ (Spaces are already unstretchable and
unshrinkable in font~|cmtt|.)
\beginlines
|\def\raggedright{\rightskip=0pt plus2em|\parbreak%
|  \spaceskip=.3333em \xspaceskip=.5em\relax}|
|\def\ttraggedright{\tt\rightskip=0pt plus2em\relax}|
\endlines

Now we come to special symbols and accents, which depend primarily on
the characters available in the Computer Modern fonts. Different
constructions will be necessary if other styles of type are used.
When a symbol is built up by forming a box, the ^|\leavevmode| macro
is called first; this starts a new paragraph, if \TeX\ is in vertical
mode, but does nothing if \TeX\ is in horizontal mode or math mode.
\beginlines
|\chardef\%=`\% \chardef\&=`\& \chardef\#=`\# \chardef\$=`\$|%
  ^^{control percent}^^{control ampersand}^^{control hash}^^{control dollar}
|\chardef|^|\ae||="1A \chardef|^|\oe||="1B \chardef|^|\o||="1C| %
  |\chardef|^|\ss||="19|
|\chardef|^|\AE||="1D \chardef|^|\OE||="1E \chardef|^|\O||="1F|
|\chardef|^|\i||="10 \chardef|^|\j||="11 % dotless letters|
|\def|^|\aa||{\accent'27a} \def|^|\l||{\char'40l}|
\smallbreak
|\def\leavevmode{\unhbox\voidb@x} % begins a paragraph, if necessary|
|\def\_{\leavevmode \kern.06em \vbox{\hrule width0.3em}}|^^{control underline}
|\def|^|\L||{\leavevmode\setbox0=\hbox{L}\hbox to\wd0{\hss\char'40L}}|
|\def|^|\AA||{\leavevmode\setbox0=\hbox{!}\dimen@=\ht0 |%
  |\advance\dimen@ by-1ex|\parbreak%
|  \rlap{\raise.67\dimen@\hbox{\char'27}}A}|
\smallbreak
|\def|^|\mathhexbox||#1#2#3{\leavevmode \hbox{$\m@th \mathchar"#1#2#3$}}|
|\def|^|\dag||{\mathhexbox279}  \def|^|\ddag||{\mathhexbox27A}|
|\def|^|\S||{\mathhexbox278} \def|^|\P||{\mathhexbox27B} |%
  |\def|^|\Orb||{\mathhexbox20D}|
\smallbreak
|\def|^|\oalign||#1{\leavevmode\vtop{\baselineskip0pt \lineskip.25ex|\parbreak%
|  \ialign{##\crcr#1\crcr}}} \def\o@lign{\lineskiplimit=0pt \oalign}|
|\def|^|\ooalign||{\lineskiplimit=-\maxdimen \oalign} % chars over each other|
|{\catcode`p=12 \catcode`t=12 \gdef\\#1pt{#1}} \let|^|\getf@ctor||=\\|
|\def\sh@ft#1{\dimen@=#1 \kern\expandafter\getf@ctor\the\fontdimen1\font|\parbreak%
|  \dimen@} % kern by #1 times the current slant|
|\def|^|\d||#1{{\o@lign{\relax#1\crcr\hidewidth\sh@ft{-1ex}.\hidewidth}}}|
|\def|^|\b||#1{{\o@lign{\relax#1\crcr\hidewidth\sh@ft{-3ex}%|\parbreak%
|    \vbox to.2ex{\hbox{\char'26}\vss}\hidewidth}}}|
|\def|^|\c||#1{{\setbox0=\hbox{#1}\ifdim\ht0=1ex \accent'30 #1%|\parbreak%
|  \else\ooalign{\unhbox0\crcr\hidewidth\char'30\hidewidth}\fi}}|
|\def|^|\copyright||{{\ooalign{\hfil\raise.07ex\hbox{c}\hfil\crcr\Orb}}}|
\smallbreak
|\def|^|\dots||{\relax\ifmmode\ldots\else$\m@th \ldots\,$\fi}|
|\def|^|\TeX||{T\kern-.1667em \lower.5ex\hbox{E}\kern-.125em X}|
\smallbreak
|\def|^|\`||#1{{\accent"12 #1}} \def|^|\'||#1{{\accent"13 #1}}|
|\def|^|\v||#1{{\accent"14 #1}} \def|^|\u||#1{{\accent"15 #1}}|
^^{esc hat}|\def|^|\=||#1{{\accent"16 #1}} \def\^#1{{\accent"5E #1}}|
|\def|^|\.||#1{{\accent"5F #1}} \def|^|\H||#1{{\accent"7D #1}}|
^^{esc tilde}|\def\~||#1{{\accent"7E #1}} \def|^|\"||#1{{\accent"7F #1}}|
|\def|^|\t||#1{{\edef\next{\the\font}\the\textfont1\accent"7F\next#1}}|
\endlines

At this point three alternative control-symbol ^{accents} are defined,
suitable for keyboards with extended character sets (cf.~Appendix~C):
\beginlines
|\let\^^_=\v       \let\^^S=\u       \let\^^D=\^|
\endlines

Various ways to fill space with ^{leaders} are provided next.
\beginlines
|\def|^|\hrulefill||{\leaders\hrule\hfill}|
|\def|^|\dotfill||{\cleaders\hbox{$\m@th \mkern1.5mu . \mkern1.5mu$}\hfill}|
|\def|^|\rightarrowfill||{$\m@th \smash- \mkern-7mu|\parbreak%
|  \cleaders\hbox{$\mkern-2mu \smash- \mkern-2mu$}\hfill|\parbreak%
|  \mkern-7mu \mathord\rightarrow$}|
|\def|^|\leftarrowfill|%
  |{$\m@th \mathord\leftarrow \mkern-7mu|\parbreak%
|  \cleaders\hbox{$\mkern-2mu \smash- \mkern-2mu$}\hfill|\parbreak%
|  \mkern-7mu \smash-$}|
\smallbreak
|\mathchardef\braceld="37A \mathchardef\bracerd="37B|
|\mathchardef\bracelu="37C \mathchardef\braceru="37D|
|\def|^|\upbracefill||{$\m@th|\parbreak%
|  \setbox0=\hbox{$\braceld$}%|\parbreak%
|  \bracelu\leaders\vrule height\ht0 depth0pt\hfill\bracerd|\parbreak%
|  \braceld\leaders\vrule height\ht0 depth0pt\hfill\braceru$}|
|\def|^|\downbracefill||{$\m@th|\parbreak%
|  \setbox0=\hbox{$\braceld$}%|\parbreak%
|  \braceld\leaders\vrule height\ht0 depth0pt\hfill\braceru|\parbreak%
|  \bracelu\leaders\vrule height\ht0 depth0pt\hfill\bracerd$}|
\endlines

\smallbreak
Finally, the fifth section of |plain.tex| closes by defining ^|\bye|:
\beginlines
^|\outer||\def\bye{\par\vfill\supereject\end} % the recommended way to stop|
\endlines

\subsection Macros for math.  The sixth section of |plain.tex| is the
longest; but it will suffice to give only excerpts here, because most
of it is simply a tedious listing of special symbols together with their
font locations, and the same information appears in Appendix~F.

Some rudimentary things come first: The control sequences ^|\sp| and ^|\sb|
are provided for people who can't easily type |^| and |_|; there are
four control symbols that provide spacing corrections; a ``^{discretionary
times sign}'' ^|\*| is defined; and then there's an interesting set of
macros that convert |f'''| into |f^{\prime\prime\prime}|:
\beginlines
|\let\sp=^ \let\sb=_        {\catcode`\_=\active \global\let_=\_}|\parbreak%
|\def|^|\,||{\mskip\thinmuskip}  \def|^|\!||{\mskip-\thinmuskip}|\parbreak%
|\def|^|\>||{\mskip\medmuskip}   \def|^|\;||{\mskip\thickmuskip}|
|\def\*{\discretionary{\thinspace\the\textfont2\char2}{}{}}|
|{\catcode`\^^Z=\active \gdef^^Z{\not=}} % ^^Z is like \ne in math|
\smallbreak
|{\catcode`\'=\active \gdef'{^\bgroup\prim@s}}|\parbreak%
|\def\prim@s{\prime\futurelet\next\pr@m@s}|\parbreak%
|\def\pr@m@s{\ifx'\next\let\nxt\pr@@@s|%
  | \else\ifx^\next\let\nxt\pr@@@t|\parbreak%
|  \else\let\nxt\egroup\fi\fi \nxt}|\parbreak%
|\def\pr@@@s#1{\prim@s} \def\pr@@@t#1#2{#2\egroup}|
\endlines

The next job is to define ^{Greek letters} and other symbols of type ^{Ord}.
Uppercase Greek letters are assigned hexadecimal codes of the form \hex{7xxx},
so that they will change families when ^|\fam| changes.
Three dots `$\,\cdots\,$' are used here and below to indicate that
additional symbols, having similar definitions, are listed in Appendix~F.
\beginlines
|\mathchardef|^|\alpha||="010B|\qquad$\cdots$\qquad|\mathchardef\omega="0121|
|\mathchardef|^|\Gamma||="7000|\qquad$\cdots$\qquad|\mathchardef\Omega="700A|
|\mathchardef|^|\aleph||="0240|\qquad$\cdots$\qquad%
  |\mathchardef\spadesuit="027F|
|\def|^|\hbar||{{\mathchar'26\mkern-9muh}}|
|\def|^|\surd||{{\mathchar"1270}}|
|\def|^|\angle||{{\vbox{\ialign{$\m@th\scriptstyle##$\crcr|\parbreak%
|      \not\mathrel{\mkern14mu}\crcr \noalign{\nointerlineskip}|\parbreak%
|      \mkern2.5mu\leaders\hrule height.34pt\hfill\mkern2.5mu\crcr}}}}|
\endlines

Large operators are assigned hexadecimal codes of the form \hex{1xxx}:
\beginlines
|\mathchardef\smallint="1273|
|\mathchardef|^|\sum||="1350|\qquad$\cdots$\qquad|\mathchardef\biguplus="1355|
|\mathchardef\intop="1352  \def|^|\int||{\intop\nolimits}|
|\mathchardef\ointop="1348 \def\oint{\ointop\nolimits}|
\endlines
Integral signs get special treatment so that their limits won't be set
above and below.

\smallskip
Binary operations are next; nothing exciting here.
\beginlines
|\mathchardef\pm="2206|\qquad$\cdots$\qquad|\mathchardef\amalg="2271|
\endlines

Relations are also fairly straightforward, except for the ones that are
constructed from other characters. The ^|\mapstochar| is a character
`$\mapstochar\mskip5mu$' of width zero that is quite useless by itself,
but it combines with right arrows to make ^|\mapsto| `$\mapsto$' and
^|\longmapsto| `$\longmapsto$'. Similarly, ^|\not| is a relation character of
width zero that puts a slash over the character that follows.  When two
relations are adjacent in a math formula, \TeX\ puts no space between them.
\beginlines
|\mathchardef\leq="3214|\qquad$\cdots$\qquad|\mathchardef\perp="323F|
|\def|^|\joinrel||{\mathrel{\mkern-3mu}}|
|\def|^|\relbar||{\mathrel{\smash-}} \def|^|\Relbar||{\mathrel=}|
|\def|^|\longrightarrow||{\relbar\joinrel\rightarrow}|\parbreak%
|\def|^|\Longrightarrow||{\Relbar\joinrel\Rightarrow}|
|\def|^|\longleftarrow||{\leftarrow\joinrel\relbar}|\parbreak%
|\def|^|\Longleftarrow||{\Leftarrow\joinrel\Relbar}|
|\def|^|\longleftrightarrow||{\leftarrow\joinrel\rightarrow}|\parbreak%
|\def|^|\Longleftrightarrow||{\Leftarrow\joinrel\Rightarrow}|
|\mathchardef\mapstochar="3237 \def\mapsto{\mapstochar\rightarrow}|\parbreak%
|\def\longmapsto{\mapstochar\longrightarrow}|
|\mathchardef\lhook="312C \def|^|\hookrightarrow||{\lhook\joinrel\rightarrow}|
|\mathchardef\rhook="312D \def|^|\hookleftarrow||{\leftarrow\joinrel\rhook}|
\smallbreak
|\def|^|\neq||{\not=}   \def|^|\models||{\mathrel|\||\joinrel=}|
|\def|^|\bowtie||{\mathrel\triangleright\joinrel\mathrel\triangleleft}|
\weakendlines

After defining characters ^|\ldotp| and ^|\cdotp| that act as math
punctuation, it is easy to define ^|\ldots| and ^|\cdots| macros that
\vadjust{\penalty-50}% fairly good breakpoint (on July 27, 1983)
give the proper spacing in most circumstances. Vertical and diagonal
dots (^|\vdots| and ^|\ddots|) are also provided here:
\beginlines
|\mathchardef\ldotp="613A\mathchardef\cdotp="6201\mathchardef\colon="603A|
|\def\ldots{\mathinner{\ldotp\ldotp\ldotp}}|
|\def\cdots{\mathinner{\cdotp\cdotp\cdotp}}|
|\def\vdots{\vbox{\baselineskip=4pt \lineskiplimit=0pt|
|    \kern6pt \hbox{.}\hbox{.}\hbox{.}}}|
|\def\ddots{\mathinner{\mkern1mu\raise7pt\vbox{\kern7pt\hbox{.}}\mkern2mu|
|    \raise4pt\hbox{.}\mkern2mu\raise1pt\hbox{.}\mkern1mu}}|
\endlines

Most of the math accents are handled entirely by the ^|\mathaccent|
primitive, but a few of the variable-width ones are constructed the
hard way:
\beginlines
|\def\acute{\mathaccent"7013 }|\qquad$\cdots$\qquad%
  |\def\ddot{\mathaccent"707F }|
|\def|^|\widetilde||{\mathaccent"0365 } \def|^|\widehat||{\mathaccent"0362 }|
|\def|^|\overrightarrow||#1{\vbox{\m@th\ialign{##\crcr|\parbreak%
|     \rightarrowfill\crcr\noalign{\kern-1pt\nointerlineskip}|\parbreak%
|     $\hfil\displaystyle{#1}\hfil$\crcr}}}|
|\def|^|\overleftarrow||#1{\vbox{\m@th\ialign{##\crcr|\parbreak%
|     \leftarrowfill\crcr\noalign{\kern-1pt\nointerlineskip}|\parbreak%
|     $\hfil\displaystyle{#1}\hfil$\crcr}}}|
|\def|^|\overbrace|%
  |#1{\mathop{\vbox{\m@th\ialign{##\crcr\noalign{\kern3pt}|\parbreak%
|     \downbracefill\crcr\noalign{\kern3pt\nointerlineskip}|\parbreak%
|     $\hfil\displaystyle{#1}\hfil$\crcr}}}|^|\limits||}|
|\def|^|\underbrace||#1{\mathop{\vtop{\m@th\ialign{##\crcr|\parbreak%
|     $\hfil\displaystyle{#1}\hfil$\crcr|%
  |\noalign{\kern3pt\nointerlineskip}|\parbreak%
|     \upbracefill\crcr\noalign{\kern3pt}}}}\limits}|
|\def|^|\skew||#1#2#3{{\muskip0=#1mu \mkern.5\muskip0|\parbreak%
|    #2{\mkern-.5\muskip0{#3}\mkern.5\muskip0}\mkern-.5\muskip0}{}}|
\endlines

Now we come to 24 delimiters that can change their size: ^^|\delimiter|
\beginlines
|\def\langle{\delimiter"426830A }      \def\rangle{\delimiter"526930B }|
|\def\lbrace{\delimiter"4266308 }      \def\rbrace{\delimiter"5267309 }|
|\def\lceil{\delimiter"4264306 }       \def\rceil{\delimiter"5265307 }|
|\def\lfloor{\delimiter"4262304 }      \def\rfloor{\delimiter"5263305 }|
|\def\lgroup{\delimiter"462833A }      \def\rgroup{\delimiter"562933B }|
|\def\lmoustache{\delimiter"437A340 }  \def\rmoustache{\delimiter"537B341 }|%
  \kern-2pt
|\def\uparrow{\delimiter"3222378 }     \def\Uparrow{\delimiter"322A37E }|
|\def\downarrow{\delimiter"3223379 }   \def\Downarrow{\delimiter"322B37F }|
|\def\updownarrow{\delimiter"326C33F } \def\arrowvert{\delimiter"026A33C }|
|\def\Updownarrow{\delimiter"326D377 } \def\Arrowvert{\delimiter"026B33D }  |
|\def\vert{\delimiter"026A30C }        \def\Vert{\delimiter"026B30D }     |
|\def\backslash{\delimiter"026E30F }   \def\bracevert{\delimiter"077C33E }|
\endlines

The `^|\big||...|^|\Bigg|' macros produce specific sizes: ^^|\Big|^^|\bigg|
\beginlines
|\def\bigl{\mathopen\big} \def\bigm{\mathrel\big} \def\bigr{\mathclose\big}|
|\def\Bigl{\mathopen\Big} \def\Bigm{\mathrel\Big} \def\Bigr{\mathclose\Big}|
|\def\biggl{\mathopen\bigg}            \def\Biggl{\mathopen\Bigg}|
|\def\biggr{\mathclose\bigg}           \def\Biggr{\mathclose\Bigg}|
|\def\biggm{\mathrel\bigg}             \def\Biggm{\mathrel\Bigg}|
|\def\big#1{{\hbox{$\left#1\vbox to 8.5pt{}\right.\n@space$}}}|
|\def\Big#1{{\hbox{$\left#1\vbox to 11.5pt{}\right.\n@space$}}}|
|\def\bigg#1{{\hbox{$\left#1\vbox to 14.5pt{}\right.\n@space$}}}|
|\def\Bigg#1{{\hbox{$\left#1\vbox to 17.5pt{}\right.\n@space$}}}|
|\def\n@space{\nulldelimiterspace=0pt \m@th}|
\endlines

There are a few other simple abbreviations related to delimiters:
\beginlines
|\def|^|\choose||{\atopwithdelims()}|
|\def|^|\brack||{\atopwithdelims[]}|
|\def|^|\brace||{\atopwithdelims\{\}}|
|\def|^|\sqrt||{\radical"270370 }|
\endlines

And now we come to something more interesting. The ^|\mathpalette| operation
constructs a formula in all four styles; it is applied here in the
implementation of\/ ^|\phantom|, ^|\smash|, ^|\root|, and other
operations. (Actually |\phantom| and |\smash| are not perfect: They
assume that the current style is uncramped.)
\beginlines
|\def\mathpalette#1#2{\mathchoice{#1\displaystyle{#2}}|\parbreak%
|  {#1\textstyle{#2}}{#1\scriptstyle{#2}}{#1\scriptscriptstyle{#2}}}|
\smallbreak
|\newbox\rootbox|
|\def\root#1\of{\setbox\rootbox=|\parbreak%
|  \hbox{$\m@th \scriptscriptstyle{#1}$}\mathpalette\r@@t}|
|\def\r@@t#1#2{\setbox0=\hbox{$\m@th #1\sqrt{#2}$}|\parbreak%
|  \dimen@=\ht0 \advance\dimen@ by-\dp0|\parbreak%
|  \mkern5mu \raise.6\dimen@\copy\rootbox \mkern-10mu \box0}|
\smallbreak
|\newif\ifv@ \newif\ifh@|
|\def|^|\vphantom||{\v@true\h@false\ph@nt}|
|\def|^|\hphantom||{\v@false\h@true\ph@nt}|
|\def\phantom{\v@true\h@true\ph@nt}|
|\def\ph@nt{\ifmmode\def\next{\mathpalette\mathph@nt}%|\parbreak%
|  \else\let\next=\makeph@nt\fi \next}|
|\def\makeph@nt#1{\setbox0=\hbox{#1}\finph@nt}|
|\def\mathph@nt#1#2{\setbox0=\hbox{$\m@th#1{#2}$}\finph@nt}|
|\def\finph@nt{\setbox2=\null \ifv@ \ht2=\ht0 \dp2=\dp0 \fi|\parbreak%
|  \ifh@ \wd2=\wd0 \fi \box2 }|
|\def\mathstrut{\vphantom(}|
\smallbreak
|\def\smash{\relax % \relax, in case this comes first in \halign|\parbreak%
|  \ifmmode\def\next{\mathpalette\mathsm@sh}\else\let\next\makesm@sh|\parbreak%
|  \fi \next}|
|\def\makesm@sh#1{\setbox0=\hbox{#1}\finsm@sh}|
|\def\mathsm@sh#1#2{\setbox0=\hbox{$\m@th#1{#2}$}\finsm@sh}|
|\def\finsm@sh{\ht0=0pt \dp0=0pt \box0 }|
\smallbreak
|\def|^|\cong||{\mathrel{\mathpalette\@vereq\sim}} % \sim over =|\parbreak%
|\def\@vereq#1#2{\lower.5pt\vbox{\lineskiplimit\maxdimen|%
 | \lineskip-.5pt|\parbreak%
|    \ialign{$\m@th#1\hfil##\hfil$\crcr#2\crcr=\crcr}}}|
|\def|^|\notin||{\mathrel{\mathpalette\c@ncel\in}}|\parbreak%
|\def\c@ncel#1#2{\m@th\ooalign{$\hfil#1\mkern1mu/\hfil$\crcr$#1#2$}}|\parbreak%
|\def|^|\rightleftharpoons||{\mathrel{\mathpalette\rlh@{}}}|\parbreak%
|\def\rlh@#1{\vcenter{\m@th\hbox{\ooalign{\raise2pt|\parbreak%
|          \hbox{$#1\rightharpoonup$}\crcr $#1\leftharpoondown$}}}}|
|\def|^|\buildrel||#1\over#2{\mathrel{\mathop{\kern0pt #2}\limits^{#1}}}|
|\def|^|\doteq||{\buildrel\textstyle.\over=}|
\endlines
These definitions illustrate how other built-up
symbol combinations could be defined to work in all four styles.

Alternate names are defined now:
\beginlines
|\let|^|\ne||=\neq        \let|^|\le||=\leq         \let|^|\ge||=\geq|
^^{esc vert}^^{esc lbrace}^^{esc rbrace}%
|\let\{=\lbrace      \let\|\||=\Vert         \let\}=\rbrace|
|\let|^|\to||=\rightarrow \let|^|\gets||=\leftarrow \let|^|\owns||=\ni|
|\let|^|\land||=\wedge    \let|^|\lor||=\vee        \let|^|\lnot||=\neg|
|\def|^|\iff||{\;\Longleftrightarrow\;}|
\endlines

The 32 common functions whose names generally appear in roman letters are listed
in Chapter~18. Only a few of the definitions need to be shown here:
^^|\arccos| ^^|\cos| ^^|\csc| ^^|\exp| ^^|\ker| ^^|\limsup| ^^|\min|
^^|\sinh| ^^|\arcsin| ^^|\cosh| ^^|\deg| ^^|\gcd| ^^|\lg| ^^|\ln| ^^|\Pr|
^^|\sup| ^^|\arctan| ^^|\cot| ^^|\det| ^^|\hom| ^^|\lim| ^^|\log| ^^|\sec|
^^|\tan| ^^|\arg| ^^|\coth| ^^|\dim| ^^|\inf| ^^|\liminf| ^^|\max|
^^|\sin| ^^|\tanh|
\beginlines
|\def\arccos{|^|\mathop||{\rm arccos}|^|\nolimits||}|
  \hskip100pt$\cdots$\qquad|\def\tanh{|^|\mathop||{\rm tanh}|^|\nolimits||}|
|\def\det{\mathop{\rm det}}|\qquad$\cdots$\qquad|\def\sup{\mathop{\rm sup}}|
|\def\liminf{\mathop{\rm lim\,inf}} \def\limsup{\mathop{\rm lim\,sup}}|
\smallskip
|\def|^|\bmod||{\nonscript\mskip-\medmuskip \mkern5mu|\parbreak%
|  |^|\mathbin||{\rm mod} \penalty900 \mkern5mu \nonscript\mskip-\medmuskip}|
|\def|^|\pmod||#1{\allowbreak \mkern18mu ({\rm mod}\,\,#1)}|
\endlines

The definition of\/ ^|\matrix| goes to some pains to ensure that two
$n$-rowed matrices will have the same height and the same depth, unless
at least one of their rows is unusually big. The definition of
^|\bordermatrix| is even more complicated, but it seems to work
reasonably well; it uses a constant ^|\p@renwd| that represents the
width of a big extensible left parenthesis.
\beginlines
|\def\matrix#1{\null\,|^|\vcenter||{\normalbaselines\m@th|\parbreak%
|    \ialign{\hfil$##$\hfil&&\quad\hfil$##$\hfil\crcr|\parbreak%
|      \mathstrut\crcr\noalign{\kern-\baselineskip}|\parbreak%
|      #1\crcr\mathstrut\crcr\noalign{\kern-\baselineskip}}}\,}|
\smallbreak
|\newdimen\p@renwd \setbox0=\hbox{\tenex B} \p@renwd=\wd0|
|\def\bordermatrix#1{\begingroup \m@th|\parbreak%
|  \setbox0=\vbox{\def\cr|%
  |{\crcr\noalign{\kern2pt\global\let\cr=\endline}}|\parbreak%
|      \ialign{$##$\hfil\kern2pt\kern\p@renwd|%
  |&\thinspace\hfil$##$\hfil|\parbreak%
|        &&\quad\hfil$##$\hfil\crcr|\parbreak%
|        \omit\strut\hfil\crcr\noalign{\kern-\baselineskip}|\parbreak%
|        #1\crcr\omit\strut\cr}}|\parbreak%
|  \setbox2=\vbox{\unvcopy0 \global\setbox1=\lastbox}|\parbreak%
|  \setbox2=\hbox{\unhbox1 \unskip \global\setbox1=\lastbox}|\parbreak%
|  \setbox2=\hbox{$\kern\wd1\kern-\p@renwd \left( \kern-\wd1|\parbreak%
|    \global\setbox1=\vbox{\box1\kern2pt}|\parbreak%
|    \vcenter{\kern-\ht1 \unvbox0 \kern-\baselineskip} \,\right)$}|\parbreak%
|  \null\;\vbox{\kern\ht1\box2}\endgroup}|
\endlines

The next macros are much simpler:
\beginlines
|\def|^|\cases||#1{\left\{\,\vcenter{\normalbaselines\m@th|\parbreak%
|    \ialign{$##\hfil$&\quad##\hfil\crcr#1\crcr}}\right.}|
|\def|^|\pmatrix||#1{\left( \matrix{#1} \right)}|
\endlines

Finally there are macros for displayed equations:
\beginlines
|\def|^|\openup||{\afterassignment\@penup\dimen@=}|\parbreak%
|\def\@penup{\advance\lineskip\dimen@|\parbreak%
|  \advance\baselineskip\dimen@ \advance\lineskiplimit\dimen@}|
|\def|^|\eqalign||#1{\null\,\vcenter{\openup1\jot \m@th|\parbreak%
|  \ialign{\strut\hfil$\displaystyle{##}$&$\displaystyle{{}##}$\hfil|\parbreak%
|      \crcr#1\crcr}}\,}|
\smallbreak
|\newif\ifdt@p|
|\def\displ@y{\global\dt@ptrue \openup1\jot \m@th|\parbreak%
|  |^|\everycr||{\noalign{\ifdt@p \global\dt@pfalse|%
  | \ifdim\prevdepth>-1000pt|\parbreak%
|      \vskip-\lineskiplimit \vskip\normallineskiplimit \fi|\parbreak%
|      \else \penalty|^|\interdisplaylinepenalty|| \fi}}}|
|\def\@lign{\tabskip=0pt\everycr={}} % restore inside \displ@y|
|\def|^|\displaylines||#1{\displ@y \tabskip=0pt|\parbreak%
|  \halign{\hbox to\displaywidth{|%
 |$\hfil\@lign\displaystyle##\hfil$}\crcr|\parbreak%
|    #1\crcr}}|
\smallbreak
|\def|^|\eqalignno||#1{\displ@y \tabskip=|^|\centering|\parbreak%
|  \halign to\displaywidth{\hfil$\@lign\displaystyle{##}$\tabskip=0pt|\parbreak%
|    &$\@lign\displaystyle{{}##}$\hfil\tabskip=\centering|\parbreak%
|    &\llap{$\@lign##$}\tabskip=0pt\crcr|\parbreak%
|    #1\crcr}}|
|\def|^|\leqalignno||#1{\displ@y \tabskip=\centering|\parbreak%
|  \halign to\displaywidth{\hfil$\@lign\displaystyle{##}$\tabskip=0pt|\parbreak%
|    &$\@lign\displaystyle{{}##}$\hfil\tabskip=\centering|\parbreak%
|    &\kern-\displaywidth\rlap{$\@lign##$}\tabskip=\displaywidth\crcr|\parbreak%
|    #1\crcr}}|
\endlines
The value of\/ |\lineskiplimit| is assumed to be |\normallineskiplimit|
plus the accumulated amount of ``opening up.'' Thus, the |\vskip|
instructions in |\displ@y| will compensate for the fact that the
first baseline of an alignment is separated by an opened-up baselineskip
from the last line preceding the display.

\subsection Macros for output. The |plain.tex| file also contains the output
routine described in Chapters 15 and~23. First there are simple
facilities related to page numbers, headings, and footings:
\beginlines
|\countdef|^|\pageno||=0 \pageno=1 % first page is number 1|
|\newtoks|^|\headline|| \headline={\hfil} % headline is normally blank|
|\newtoks|^|\footline|| \footline={\hss\tenrm\folio\hss}|\parbreak%
|  % footline is normally a centered page number in font \tenrm|
|\def|^|\folio||{\ifnum\pageno<0 \romannumeral-\pageno \else\number\pageno \fi}|
|\def|^|\nopagenumbers||{\footline={\hfil}} % blank out the footline|
|\def|^|\advancepageno||{\ifnum\pageno<0 \global\advance\pageno by -1|\parbreak%
|  \else\global\advance\pageno by 1 \fi} % increase |\||pageno|\|
\smallbreak
|\newif\ifr@ggedbottom|
|\def|^|\raggedbottom||{\topskip10pt plus60pt \r@ggedbottomtrue}|
|\def|^|\normalbottom||{\topskip10pt \r@ggedbottomfalse} % undoes \raggedbottom|
\endlines

The ^|\footnote| macro has a few subtle features that can best be appreciated
by someone who reads Chapter~15 very carefully. It also uses some
^|\bgroup| and ^|\futurelet| and ^|\aftergroup| trickery, so that the footnote
text does not need to be a ^{parameter} to |\vfootnote|:
\beginlines
^|\newinsert||\footins|
|\def\footnote#1{\let\@sf=\empty|%
 | % parameter #2 (the text) is read later|\parbreak%
|  \ifhmode\edef\@sf{\spacefactor=\the\spacefactor}\/\fi|\parbreak%
|  #1\@sf\vfootnote{#1}}|
|\def|^|\vfootnote||#1{\insert\footins\bgroup|\parbreak%
|  \interlinepenalty=\interfootnotelinepenalty|\parbreak%
|  \splittopskip=\ht\strutbox % top baseline for broken footnotes|\parbreak%
|  \splitmaxdepth=\dp\strutbox \floatingpenalty=20000|\parbreak%
|  \leftskip=0pt \rightskip=0pt \spaceskip=0pt \xspaceskip=0pt|\parbreak%
|  \textindent{#1}\footstrut\futurelet\next\fo@t}|
|\def\fo@t{\ifcat\bgroup\noexpand\next \let\next\f@@t|\parbreak%
|  \else\let\next\f@t\fi \next}|
|\def\f@@t{\bgroup\aftergroup\@foot\let\next}|
|\def\f@t#1{#1\@foot}|
|\def\@foot{\strut\egroup}|
|\def\footstrut{\vbox to\splittopskip{}}|
|\skip\footins=\bigskipamount % space added when footnote is present|
|\count\footins=1000 % footnote magnification factor (1 to 1)|
|\dimen\footins=8in % maximum footnotes per page|
\endlines

^{Floating insertions} are handled by doing an ^|\insert| whose
vertical list consists of a penalty item followed by a single box:
\beginlines
|\newinsert\topins \newif\ifp@ge \newif\if@mid|
|\def|^|\topinsert||{\@midfalse\p@gefalse\@ins}|
|\def|^|\midinsert||{\@midtrue\@ins}|
|\def|^|\pageinsert||{\@midfalse\p@getrue\@ins}|
|\skip\topins=0pt % no space added when a topinsert is present|
|\count\topins=1000 % magnification factor (1 to 1)|
|\dimen\topins=\maxdimen % no limit per page|
|\def\@ins{\par\begingroup\setbox0=\vbox\bgroup} % start a \vbox|
|\def\endinsert{\egroup % finish the \vbox|\parbreak%
|  \if@mid \dimen@=\ht0 \advance\dimen@ by\dp\z@ |%
 |\advance\dimen@ by12\p@|\parbreak%
|    \advance\dimen@ by\pagetotal \advance\dimen@ by-\pageshrink|\parbreak%
|    \ifdim\dimen@>\pagegoal \@midfalse\p@gefalse\fi\fi|\parbreak%
|  \if@mid \bigskip \box0 \bigbreak|\parbreak%
|  \else\insert\topins{\penalty100 % floating insertion|\parbreak%
|    |^|\splittopskip||=0pt |^|\splitmaxdepth||=\maxdimen |%
  ^|\floatingpenalty||=0|\parbreak%
|    \ifp@ge \dimen@=\dp0|\parbreak%
|     \vbox to\vsize{\unvbox0 \kern-\dimen@} % depth is zero|\parbreak%
|    \else \box0 \nobreak\bigskip\fi}\fi\endgroup}|
\endlines

Most of the ^|\output| routine appears in Chapter 23; it is given here
in full:
\beginlines
|\output={\plainoutput}|
|\def|^|\plainoutput|%
  |{\shipout\vbox{\makeheadline\pagebody\makefootline}%|\parbreak%
|  \advancepageno|\parbreak%
|  \ifnum\outputpenalty>-20000 \else\dosupereject\fi}|
|\def|^|\pagebody||{\vbox to\vsize{\boxmaxdepth=\maxdepth \pagecontents}}|
|\def|^|\makeheadline||{\vbox to 0pt{\vskip-22.5pt|\parbreak%
|    \line{\vbox to8.5pt{}\the\headline}\vss}\nointerlineskip}|
|\def|^|\makefootline||{\baselineskip=24pt \lineskiplimit=0pt|\parbreak%
|    \line{\the\footline}}|
|\def|^|\dosupereject|%
  |{\ifnum\insertpenalties>0 % something is being held over|\parbreak%
|  \line{}\kern-\topskip\nobreak\vfill\supereject\fi}|
\smallbreak
|\def|^|\pagecontents||{\ifvoid\topins\else\unvbox\topins\fi|\parbreak%
|  \dimen@=\dp255 \unvbox255|\parbreak%
|  \ifvoid\footins\else % footnote info is present|\parbreak%
|    \vskip\skip\footins \footnoterule \unvbox\footins\fi|\parbreak%
|  \ifr@ggedbottom \kern-\dimen@ \vfil \fi}|
|\def|^|\footnoterule||{\kern-3pt|\parbreak%
|  \hrule width 2truein \kern 2.6pt} % the \hrule is .4pt high|
\endlines

\subsection Hyphenation and everything else.  The last part of |plain.tex|
reads the hyphenation patterns and exceptions found on file ^|hyphen.tex|
(see Appendix~H\null); then it defines a few miscellaneous macros, sets
up ^|\rm| type, and that's all!
\beginlines
^|\lefthyphenmin||=2 |^|\righthyphenmin||=3 % disallow x- or -xx breaks|
|\input hyphen % the hyphenation patterns and exceptions|
\nobreak\smallskip
|\def|^|\magnification||{|^|\afterassignment||\m@g\count@}|
|\def\m@g{\mag=\count@|\parbreak%
|  \hsize6.5truein\vsize8.9truein\dimen\footins8truein}|
\smallbreak
|\def|^|\loggingall|%
  |{\tracingcommands=2 \tracingstats=2|\parbreak%
|  \tracingpages=1 \tracingoutput=1 \tracinglostchars=1 |\parbreak%
|  \tracingmacros=2 \tracingparagraphs=1 \tracingrestores=1 |\parbreak%
|  \showboxbreadth=\maxdimen \showboxdepth=\maxdimen}|
|\def|^|\tracingall||{\tracingonline=1 \loggingall}|
\smallbreak
|\def|^|\showhyphens|%
  |#1{\setbox0=\vbox{\parfillskip0pt \hsize=\maxdimen \tenrm|\parbreak%
|  \pretolerance=-1 \tolerance=-1 \hbadness=0 \showboxdepth=0 \ #1}}|
\smallbreak
|\normalbaselines\rm % select roman font|
|\nonfrenchspacing % punctuation affects the spacing|
|\catcode`@=12 % at signs are no longer letters|
\nobreak\smallskip
|\def|^|\fmtname||{plain}|
|\def\fmtversion{3.1415926535} % identifies the current format|
\endlines
The format name and version number are recorded in control sequences,
in order to help the people who might have to explain why something
doesn't work. Macro files like |plain.tex| should not be changed in any way,
except with respect to preloaded fonts, unless the changes are authorized
by the author of the macros.

\endchapter

The purpose of a programming system is to make a computer easy to use.
To do this, it furnishes languages and various facilities
that are in fact programs invoked and controlled by language features.
But these facilities are bought at a price:
the external description of a programming system is ten to twenty times
as large as the external description of the computer system itself.
The user finds it far easier to specify any particular function,
but there are far more to choose from,
and far more options and formats to remember.
\author FREDERICK P. ^{BROOKS}, JR., {\sl The Mythical Man Month\/} (1975)
  % p 43

\bigskip

When someone says, ``I want a programming language
in which I need only say what I wish done,''
give him a lollipop.
\author ALAN ^{PERLIS}, {\sl Epigrams on Programming\/} (1982)
  % SIGPLAN Notices 17,9 (September 82), 7--13.
  % There are many more, like "Editing is a rewording activity."

\eject
\beginchapter Appendix C. Character\\Codes

\ninepoint
Different computers tend to have different ways of representing the
characters in files of text, but \TeX\ gives the same results on
all machines, because it converts everything to a standard internal
code when it reads a file. \TeX\ also converts back from its internal
representation to the appropriate external code, when it writes
a file of text; therefore most users need not be aware of the fact
that the ^{codes} have actually switched back and forth inside the machine.

The purpose of this appendix is to define \TeX's internal code,
which has the same characteristics on all implementations of \TeX.
The existence of such a code is important, because it
makes \TeX\ constructions ``portable.'' For example, \TeX\ allows
^{alphabetic constants} like |`b| to be used as numbers; the fact
that |`b| always denotes the integer~98 means that we can
write machine-independent macros that decide, for instance, whether
a given character is a digit between |0| and |9|.
Furthermore the internal code of \TeX\ also survives in its ^|dvi|
output files, which can be printed by software that knows nothing
about where the |dvi| data originated; essentially the same
output will be obtained from all implementations of \TeX, regardless
of the host computer, because the |dvi| data is expressed in a
machine-independent code.

\smallskip
\TeX's internal code is based on the American Standard Code for
Information Interchange, known popularly as ``^{ASCII}.'' There are
128 codes, numbered 0~to~127; we conventionally express the numbers
in octal notation, from \oct{000} to \oct{177}, or in
hexadecimal notation, from \hex{00} to \hex{7F}. Thus, the value of
|`b| is normally called \oct{142} or \hex{62}, not 98. In the
ASCII scheme, codes \oct{000} through \oct{040} and
code \oct{177} are~assigned to special functions; for example,
code \oct{007} is called |BEL|, and it means ``Ring the bell.''
The other 94 codes are assigned to visible symbols. Here is a
chart that shows ASCII codes in such a way that octal and hexadecimal
equivalents can easily be read off:
\beginchart{\global\count@='41\tentt
  \def\chartstrut{\lower4.3pt\vbox to13.6pt{}}}
&\oct{00x}&&NUL&&SOH&&STX&&ETX&&EOT&&ENQ&&ACK&&BEL&&\oddline0
&\oct{01x}&&BS&&HT&&LF&&VT&&FF&&CR&&SO&&SI&\evenline
&\oct{02x}&&DLE&&DC1&&DC2&&DC3&&DC4&&NAK&&SYN&&ETB&&\oddline1
&\oct{03x}&&CAN&&EM&&SUB&&ESC&&FS&&GS&&RS&&US&\evenline
&\oct{04x}&&SP&&\:&&\:&&\:&&\:&&\:&&\:&&\:&&\oddline2
&\oct{05x}&&\:&&\:&&\:&&\:&&\:&&\:&&\:&&\:&\evenline
&\oct{06x}&&\:&&\:&&\:&&\:&&\:&&\:&&\:&&\:&&\oddline3
&\oct{07x}&&\:&&\:&&\:&&\:&&\:&&\:&&\:&&\:&\evenline
&\oct{10x}&&\:&&\:&&\:&&\:&&\:&&\:&&\:&&\:&&\oddline4
&\oct{11x}&&\:&&\:&&\:&&\:&&\:&&\:&&\:&&\:&\evenline
&\oct{12x}&&\:&&\:&&\:&&\:&&\:&&\:&&\:&&\:&&\oddline5
&\oct{13x}&&\:&&\:&&\:&&\:&&\:&&\:&&\:&&\:&\evenline
&\oct{14x}&&\:&&\:&&\:&&\:&&\:&&\:&&\:&&\:&&\oddline6
&\oct{15x}&&\:&&\:&&\:&&\:&&\:&&\:&&\:&&\:&\evenline
&\oct{16x}&&\:&&\:&&\:&&\:&&\:&&\:&&\:&&\:&&\oddline7
&\oct{17x}&&\:&&\:&&\:&&\:&&\:&&\:&&\:&&DEL&\evenline
\endchart

Ever since ASCII was established in the early 1960s, people have had
different ideas about what to do with positions \oct{000}--\oct{037} and
\oct{177}, because most of the functions assigned to those codes are
appropriate only for special purposes like file transmission, not for
applications to printing or to interactive computing. It turned out that
manufacturers soon started producing line printers that were capable of
generating 128 characters, 33~of~which were tailored to the special needs
of particular customers; part of the advantage of a standard code was
therefore lost. On the other hand, the remaining 95~codes (including
\oct{40}=|SP|, a blank space) have become widely adopted, and they are now
implanted within most of today's computer terminals. When an ASCII
keyboard is available, you can specify each of the 128 codes to \TeX\ in
terms of the 95 standard characters, as follows:
\beginchart{\def\?{\char`^\char`^\:}\global\count@='100
  \postdisplaypenalty=0 \tentt}
&\oct{00x}&&\?&&\?&&\?&&\?&&\?&&\?&&\?&&\?&&\oddline0
&\oct{01x}&&\?&&\?&&\?&&\?&&\?&&\?&&\?&&\?&\evenline
&\oct{02x}&&\?&&\?&&\?&&\?&&\?&&\?&&\?&&\?&&\oddline1
&\oct{03x}&&\?&&\?&&\?&&\?&&\?&&\?&&\?&&\?&\evenline
&\global\count@='40
\oct{04x}&&\:&&\:&&\:&&\:&&\:&&\:&&\:&&\:&&\oddline2
&\oct{05x}&&\:&&\:&&\:&&\:&&\:&&\:&&\:&&\:&\evenline
&\oct{06x}&&\:&&\:&&\:&&\:&&\:&&\:&&\:&&\:&&\oddline3
&\oct{07x}&&\:&&\:&&\:&&\:&&\:&&\:&&\:&&\:&\evenline
&\oct{10x}&&\:&&\:&&\:&&\:&&\:&&\:&&\:&&\:&&\oddline4
&\oct{11x}&&\:&&\:&&\:&&\:&&\:&&\:&&\:&&\:&\evenline
&\oct{12x}&&\:&&\:&&\:&&\:&&\:&&\:&&\:&&\:&&\oddline5
&\oct{13x}&&\:&&\:&&\:&&\:&&\:&&\:&&\:&&\:&\evenline
&\oct{14x}&&\:&&\:&&\:&&\:&&\:&&\:&&\:&&\:&&\oddline6
&\oct{15x}&&\:&&\:&&\:&&\:&&\:&&\:&&\:&&\:&\evenline
&\oct{16x}&&\:&&\:&&\:&&\:&&\:&&\:&&\:&&\:&&\oddline7
&\oct{17x}&&\:&&\:&&\:&&\:&&\:&&\:&&\:&&\char`^\char`^?&\evenline
\endchart
(Here `|^^|' ^^{uparrow uparrow} doesn't necessarily mean two
circumflex characters; it means two identical characters whose
current |\catcode| is~7. In such cases \TeX\ simply adds or
subtracts \oct{100} from the internal code of the character
that immediately follows. For example, |*|~can also be typed as |^^j|;
|j|~can also be typed as |^^*|.)

An extended ASCII code intended for text editing and interactive computing
 was developed at several universities about 1965, and
for many years there have been terminals in use at Stanford, MIT,
Carnegie-Mellon, and elsewhere that have 120 or~121 symbols, not just~95.
Aficionados of these keyboards (like the author of this book)
are loath to give up their extra characters; it seems that
such people make heavy use of about 5~of the extra~25, and occasional
use of the other~20, although
different people have different groups of five. For example, the
author developed \TeX\ on a keyboard that includes the symbols
{\tentex\char'30},~{\tentex\char1}, {\tentex\char'32}, {\tentex\char'34},
and~{\tentex\char'35}, and he finds that this makes it much more pleasant
to type class notes,
technical papers, and computer programs of the kind he likes to write;
his logician friends make heavy use of the {\tentex\char'24} and
{\tentex\char'25} keys; and so on. It is recommended that \TeX\
implementations on systems with large character sets be consistent with the
following codes:
\beginchart{\tentex\postdisplaypenalty=0}
\normalchart
\endchart
Of course, designers of \TeX\ macro packages that are intended to be
widely used should stick to the standard ASCII characters.

Incidentally, the ASCII character |^| that appears in position \oct{136}
is sometimes called a ``^{caret},'' but dictionaries of English tell us
that a caret is a larger symbol, more like character \oct{004} in the
extended set above. The correct name for |^| is ``^{circumflex},'' but
this is quite a mouthful, so a shorter name like ``^{hat}'' is preferable.
It seems desirable to preserve the traditional distinction between
caret and hat.

The extended code shown above was developed at MIT; it is similar to, but
slightly better than, the code implemented at Stanford. Seven of the codes
are conventionally assigned to the standard ASCII control functions
|NUL| (^\<null>), |HT| (^\<tab>), |LF| (^\<linefeed>), |FF|
(^\<formfeed>), |CR| (^\<return>), |ESC| (^\<escape>), and
|DEL| (^\<delete>), and they appear in the standard ASCII positions; hence the
corresponding seven characters
{\tentex\char0~\char'11~\char'12~\char'14~\char'15~\char'33~\char'177}
do not actually appear on the keyboard. These seven ``hidden''
characters show up only on certain output devices.

Modern keyboards allow 256 codes to be input, not just 128; so \TeX\
represents characters internally as numbers in the range 0--255 (i.e.,
\oct{000}--\oct{377}, or \hex{00}--\hex{FF}). Implementations of \TeX\
differ in which characters they will accept in input files and which
they will transmit to output files; these subsets can be specified
independently. A completely permissive version of \TeX\ allows full
256-character input and output; other versions might ignore all
but the visible characters of ASCII; still other versions might
distinguish the tab character (code \oct{011}) from a space on input,
but might output each tab as a sequence of three characters |^^I|.

Many people, unfortunately, have the opposite problem:\ Instead of the
95~standard characters and some others, they have fewer than~95 symbols
actually available. What can be done in such cases? Well, it's possible
to use \TeX\ with fewer symbols, by invoking more control
sequences; for example, plain \TeX\ defines ^|\lq|, ^|\rq|, ^|\lbrack|,
^|\rbrack|, ^|\sp|, and ^|\sb|, so that you need not type |`|,~|'|, |[|, |]|,
|^|, and~|_|, respectively.

A person who implements \TeX\ on computer systems that do not have
95~externally representable symbols should adhere to the following
guidelines: \ (a)~Stay as close as possible to the ASCII conventions.
\ (b)~Make sure that codes \oct{041}--\oct{046}, \oct{060}--\oct{071},
\oct{136}, \oct{141}--\oct{146}, and \oct{160}--\oct{171} are present and that
each unrepresentable
internal code $<\null$\oct{200} leads to a representable code when \oct{100} is
added or subtracted; then all 256 codes can be input and output.
\ (c)~Cooperate with everyone else who shares
the same constraints, so that you all adopt the same policy.
\ (See Appendix~J for information about the \TeX\ Users Group.)

\def\Russiantt#1{{\tt\hbox to.5em{\hss\eighttt\char#1\hss}}}
Very few conventions about character codes are hardwired into \TeX:
Almost everything can be changed by a format package that changes parameters
like |\escapechar| and sets up the |\catcode|, |\mathcode|, |\uccode|,
|\lccode|, |\sfcode|, and |\delcode| tables.
Thus a \TeX\ manuscript that has been written in ^{Denmark}, say,
can be run in California, and vice versa, even though quite different
conventions might be used in different countries. The only character codes
that \TeX\ actually ``knows'' are these: \ (1)~|INITEX| initializes
the code tables as described in Appendix~B\null; the same initialization is
done by all implementations of \TeX. \ (2)~\TeX\ uses the character codes
\]|+-.,`'"<=>0123456789ABCDEF| in its syntax rules (Chapters 20, 24, and
Appendix~H\null), and it
uses most of the uppercase and lowercase letters in its ^{keywords}
|pt|, |to|, |plus|, etc. These same codes and keywords are used in
all implementations of \TeX\null. For example, if \TeX\ is implemented for
a purely ^{Cyrillic} ^^{Russian} keyboard, the letter `\Russiantt5' should be
assigned to code \oct{160} and `\Russiantt{`T}' to code \oct{164}, so that
`\Russiantt5\Russiantt{`T}' still means `|pt|'; or else control sequences
should be defined so that what \TeX\ sees is equivalent to the keywords
it needs. \ (3)~The operations |\number|,
|\romannumeral|, |\the|, and |\meaning| can generate letters, digits, spaces,
decimal points, minus signs, double quotes, colons, and `|>|' signs; these same
codes are generated in all implementations of \TeX.
\ (4)~The |\hyphenation| and |\pattern| commands described in Appendix~H
give special interpretation to the ten digits and to the characters
`|.|'\ and~`|-|'. \ (5)~The codes for the four characters |$|~|.|~|{|~|}|
are inserted when \TeX\ recovers from certain errors, and braces are
inserted around an ^|\output| routine; appropriate catcodes are attached
to these tokens, so it doesn't matter if these symbols have their plain
\TeX\ meanings or not. \ (6)~There is a special convention for
representing characters 0--255 in the hexadecimal forms
|^^00|--|^^ff|, explained in Chapter~8. This convention is always
acceptable as input, when |^| is any character of catcode~7. Text
output is produced with this convention only when representing
characters of code $\ge128$ that a \TeX\ installer has chosen not to
output directly.

\endchapter

Code sets obtained by modifying the standard as shown above
or by other replacements are nonstandard.
\author ASA SUBCOMMITTEE X3.2, {\sl American Standard\break %
  Code for Information Interchange\/^^{ASCII}} (1963)
% in {\sl Communications of the ACM\/}

\bigskip

Both the Stanford and DEC uses of the ASCII control characters
are in violation of the USA Standard Code,
but no Federal Marshal is likely to come running out
and arrest people who type control-T to their computers.
\author BRIAN ^{REID}, {\sl SCRIBE Introductory User's Manual\/} (1978) % p82

\eject
\beginchapter Appendix D. Dirty Tricks

\TeX\ was designed to do the ordinary tasks of typesetting: to make paragraphs
and pages. But the underlying mechanisms that facilitate ordinary
typesetting---e.g., boxes, glue, penalties, and macros---are extremely
versatile; hence people have discovered sneaky ways to coerce \TeX\ into
doing tricks quite different from what its author originally had in mind.
Such clever constructions are not generally regarded as examples of
``high \TeX''; but many of them have turned out to be useful and
instructive, worthy of being known (at least by a few wizards). The purpose
of this appendix is to introduce crafty and/or courageous readers
to the nether world of \TeX arcana.

\ninepoint\medskip
\setbox0=\hbox spread-6\fontdimen4\font % that removes all the shrinkability
 {\strut Please don't read this material until you've}
\setbox1=\hbox to\wd0{\strut had plenty of experience with plain \TeX.}
\line{\leaders\hbox{\dbend\kern2pt}\hfil\vtop{\box0\box1}}
\nointerlineskip
\noindent\strut After you have read and understood the secrets below, you'll
know all sorts of devious combinations of \TeX\ commands,
and you will often be tempted to write inscrutable macros. Always remember,
however, that there's usually a simpler and better way to do something
than the first way that pops into your head. You may not have to
resort to any subterfuge at all, since \TeX\ is able to do lots of things
in a straightforward way. Try for simple solutions first.

\subsection Macro madness. If you need to write complicated ^{macros}, you'll
need to be familiar with the fine points in Chapter~20. \TeX's control
sequences are divided into two main categories, ``expandable'' and
``unexpandable''; the former category includes all macros and |\if...\fi|
tests, as well as special operations like |\the| and |\input|, while the
latter category includes the primitive commands listed in Chapters~24--26.
The expansion of expandable tokens takes place in \TeX's ``^{mouth},''
but primitive commands (including assignments) are done in \TeX's
``^{stomach}.'' One important consequence of this structure is that
it is impossible to redefine a control sequence or to advance a register
while \TeX\ is expanding the token list of, say, a |\message| or |\write|
command; assignment operations are done only when \TeX\ is building a
vertical or horizontal or math list.\looseness=-1

For example, it's possible to put |\n| asterisks into a paragraph,
by saying simply `|{|^|\loop||\ifnum\n>0 *\advance\n-1 \repeat}|'.
But it's much more difficult to define a
control sequence |\asts| to consist of exactly |\n| consecutive asterisks.
If\/ |\n| were known to be at most~5, say, it would be possible to write
\begintt
\edef\asts{\ifcase\n\or*\or**\or***\or****\or*****\else\bad\fi}
\endtt
since \TeX\ handles |\ifcase| in its mouth. But for general |\n| it would
be impossible to use a construction like
`|\edef\asts{\loop\ifnum\n>0 *\advance\n-1 \repeat}|', since |\n| doesn't change
during an ^|\edef|. A more elaborate program is needed; e.g.,
\begintt
{\xdef\asts{}
  \loop\ifnum\n>0 \xdef\asts{\asts*}\advance\n-1 \repeat}
\endtt
And here's another solution (which may be faster, because
^{token list registers} can be expanded more quickly than macros
on some implementations, using ^|\the|):
\begintt
\newcount\m \newtoks\t \m=\n \t={}
\loop \ifnum\m>0 \t=\expandafter{\the\t*}\advance\m-1 \repeat
\edef\asts{\the\t}
\endtt

However, both of these solutions have a running time proportional to the
square of\/~|\n|. There's a much quicker way to do the job:
\begintt
\begingroup\aftergroup\def\aftergroup\asts\aftergroup{
\loop \ifnum\n>0 \aftergroup*\advance\n-1 \repeat
\aftergroup}\endgroup
\endtt
Get it? The ^|\aftergroup| commands cause a whole list of other tokens to be
saved up for after the group! This method has only one flaw, namely
that it takes up |\n| cells of space on \TeX's ^{input stack} and |\n|
more on \TeX's ^{save stack}; hence a special version of \TeX\ may be
required when |\n| is larger than 150 or~so.

(Incidentally, there's a completely different way to put |\n| asterisks
into a paragraph, namely to say `|\setbox0=\hbox{*}|%
^|\cleaders|^|\copy||0\hskip\n\wd0|'. This may seem to be the fastest
solution of all; but actually it is not so fast, when all things are
considered, since it generates four bytes of ^|dvi| output per asterisk,
compared to only one byte per asterisk in the other methods. Input/output
time takes longer than computation time, both in \TeX\ itself and in
the later stages of the printing process.)

The problem just solved may seem like a rather special application; after
all, who needs a control sequence that contains a variable number of
asterisks? But the same principles apply in other similar cases,
e.g., when you want to construct a variable-length ^|\parshape|
specification. Similarly, many of the ``toy problems'' solved below are
meant to illustrate paradigms that can be used in real-life situations.

The precise rules for expansion are explained in Chapter~20; and the
best way to get familiar with \TeX's expansion mechanism is to watch it
in action, looking at the log file when |\tracingmacros=2| and
|\tracingcommands=2|. One of the important ways to change the normal
order of expansion is to use ^|\expandafter|; the construction
\begintt
\expandafter\a\b
\endtt
causes |\b| to be expanded first, then~|\a|.
And since |\expandafter| is itself expandable, the construction
\begintt
\expandafter\expandafter\expandafter\a\expandafter\b\c
\endtt
causes |\c| to be expanded first, then |\b|, then |\a|. \ (The next step,
\begintt
\expandafter\expandafter\expandafter\expandafter
 \expandafter\expandafter\expandafter\a
 \expandafter\expandafter\expandafter\b\expandafter\c\d
\endtt
is probably too lengthy to be of any use.)

It's possible to make good use of\/ |\expandafter\a\b| even when |\a|
isn't expandable. For example, the token list assignment
`|\t=\expandafter{\the\t|~|*}|' in the example on the previous page
was able to invade territory where expansion is normally suppressed,
by expanding after a left brace. Similarly,
\begintt
\t=\expandafter{\expandafter*\the\t}
\endtt
would have worked; and ^^|\uppercase|
\begintt
\uppercase\expandafter{\romannumeral\n}
\endtt
yields the value of register |\n| in ^{uppercase roman numerals}.
^^{roman numerals, uppercase}

Here's a more interesting example: Recall that ^|\fontdimen||1| is the
amount of ``^{slant} per point'' of a font; hence, for example,
`^|\the||\fontdimen1\tenit|' expands to `{\tt\the\fontdimen1\tenit}',
where the characters `|pt|' are of category~12. After the macro definitions
\begintt
{\catcode`p=12 \catcode`t=12 \gdef\\#1pt{#1}}
\let\getfactor=\\
\def\kslant#1{\kern\expandafter\getfactor\the\fontdimen1#1\ht0 }
\endtt
one can write, e.g., `|\kslant\tenit|' and this will expand to
`|\kern0.25\ht0|'. If the boundary of\/ |\box0| is considered to be
slanted by 0.25 horizontal units per vertical unit, this kern measures
the horizontal distance by which the top edge of the box is skewed with
respect to an edge at the baseline. All of the computation of\/
|\kslant| is done in \TeX's mouth; thus, the mouth can do some rather
complicated things even though it cannot assign new values. \ (Incidentally,
an indirect method was used here to define the control sequence ^|\getfactor|
when the character~|t| had category~12, since control words normally consist
only of letters. The alternative construction
\begintt
{\catcode`p=12 \catcode`t=12
  \csname expandafter\endcsname\gdef
  \csname getfactor\endcsname#1pt{#1}}
\endtt
would also have worked, since ^|\csname| and ^|\endcsname| don't
contain `|p|' or `|t|'!)

The mechanism by which \TeX\ determines the ^{arguments} of a macro
can be applied in unexpected ways. Suppose, for example, that |\t|~is
a token list register that contains some text; we wish to determine if
at least one asterisk (|*|$_{12}$) appears in that text. Here's
one way to do it:
\begintt
\newif\ifresult % for the result of a computed test
\def\atest#1{\expandafter\a\the#1*\atest\a}
\long\def\a#1*#2#3\a{\ifx\atest#2\resultfalse\else\resulttrue\fi}
\endtt
^^|\long|^^|\newif|
Now after `|\atest\t|', the control sequence |\ifresult|
will be |\iftrue| or |\iffalse|, depending on whether or not |\t|~contains
an asterisk. \ (Do you see why?) \ And here's a slightly more elegant way
to do the same thing, using ^|\futurelet| to look ahead:
\begintt
\def\btest#1{\expandafter\b\the#1*\bb}
\long\def\b#1*{\futurelet\next\bb}
\long\def\bb#1\bb{\ifx\bb\next\resultfalse\else\resulttrue\fi}
\endtt
In both cases the solution works if\/ |\t| contains control sequence tokens
as well as character tokens, provided that the special control sequences
|\atest|, |\a|, and |\bb| don't appear. Notice, however, that an asterisk
is ``hidden'' if it appears within a ^{group}~|{...}|; the test is limited
to asterisks at nesting level zero. A token list register is always balanced
with respect to grouping, so there is no danger of the test leading
to error messages concerning missing braces or extra braces.

We can apply the ideas in the preceding paragraph to solve a problem
related to generalized math formatting: ^^{displays, non-centered}
The goal is to set \TeX\ up so that the respective constructions
`\thinspace|$$|$\,\alpha\,$|$$|\thinspace', ^^{dollar dollar}
`\thinspace|$$|$\,\alpha\,$^|\eqno|$\,\beta\,$|$$|\thinspace', and
`\thinspace|$$|$\,\alpha\,$^|\leqno|$\,\beta\,$|$$|\thinspace' will cause
a macro |$$\generaldisplay$$| to be invoked, with |\eq| defined to be $\alpha$.
^^{communication between macros}
Furthermore, when an equation number~$\beta$ is present, it should be stored
in |\eqn|, and the test |\ifeqno| should be true.
In such cases |\ifleqno| should distinguish |\leqno| from |\eqno|.
Here $\alpha$~and~$\beta$ are arbitrary balanced token lists that don't
contain either |\eqno| or |\leqno| at nesting level zero. The
following macros do the required maneuvers:
\begintt
\newif\ifeqno \newif\ifleqno \everydisplay{\displaysetup}
\def\displaysetup#1$${\displaytest#1\eqno\eqno\displaytest}
\def\displaytest#1\eqno#2\eqno#3\displaytest{%
  \if!#3!\ldisplaytest#1\leqno\leqno\ldisplaytest
  \else\eqnotrue\leqnofalse\def\eqn{#2}\def\eq{#1}\fi
  \generaldisplay$$}
\def\ldisplaytest#1\leqno#2\leqno#3\ldisplaytest{\def\eq{#1}%
  \if!#3!\eqnofalse\else\eqnotrue\leqnotrue\def\eqn{#2}\fi}
\endtt
An examination of the three cases
|$$|$\,\alpha\,$|$$|,
|$$|$\,\alpha$|\eqno|$\,\beta\,$|$$|,
|$$|$\,\alpha$|\leqno|$\,\beta\,$|$$| shows that the correct actions will
ensue. Parameter |#3| in the tests `|\if!#3!|'\ will be either empty or
|\eqno| or |\leqno|; thus, the condition will be false (and the second
`|!|'\ will be skipped) unless |#3| is empty.

Returning to the problem of |*|'s in |\t|, suppose that it's
necessary to consider |*|'s at all levels of nesting. Then a slower
routine must be used:
\begintt
\def\ctest#1{\resultfalse\expandafter\c\the#1\ctest}
\def\c{\afterassignment\cc\let\next= }
\def\cc{\ifx\next\ctest \let\next\relax
  \else\ifx\next*\resulttrue\fi\let\next\c\fi \next}
\endtt
Here ^|\afterassignment| has been used to retain control after a
non-future ^|\let|; the `|= |' ensures that exactly one token is ^^{equals}
swallowed per use of\/~|\c|. This routine could be modified in an obvious
way to count the total number of |*|'s and/or tokens in~|\t|. Notice the
`|\let\next|' instructions in~|\cc|; it should be clear why the alternative
\begintt
\def\cc{\ifx\next\ctest\else\ifx\next*\resulttrue\fi\c\fi}
\endtt
would not work. \ (The latter |\c| would always swallow a `|\fi|'.)

^{Space tokens} are sometimes anomalous, so they deserve special care. The
following macro ^|\futurenonspacelet| behaves essentially like
|\futurelet| except that it discards any implicit or explicit space tokens
that intervene before a nonspace is scanned:
\begintt
\def\futurenonspacelet#1{\def\cs{#1}%
  \afterassignment\stepone\let\nexttoken= }
\def\\{\let\stoken= } \\ % now \stoken is a space token
\def\stepone{\expandafter\futurelet\cs\steptwo}
\def\steptwo{\expandafter\ifx\cs\stoken\let\next=\stepthree
  \else\let\next=\nexttoken\fi \next}
\def\stepthree{\afterassignment\stepone\let\next= }
\endtt
An operation like |\futurenonspacelet| is useful, for example,
when implementing macros that have a variable number of arguments.

Notice that `|\def\stepthree#1{\stepone}|' would not work here, because of
\TeX's rule that a \]$_{10}$ token is bypassed if it would otherwise be treated
as an ^{undelimited} ^{argument}. Because of this rule it is
difficult to distinguish explicit space tokens from implicit ones.
The situation is surprisingly complex, because it's possible to
use ^|\uppercase| to create ``^{funny space}'' tokens like |*|$_{10}$;
^^|\uccode| for example, the commands
\begintt
\uccode` =`* \uppercase{\uppercase{\def\fspace{ }\let\ftoken= } }
\endtt
make |\fspace| a macro that expands to a funny space, and they make
|\ftoken| an implicit funny space. \ (The tests ^|\if||\fspace*|,
|\if\ftoken*|, ^|\ifcat||\fspace\stoken|, and |\ifcat\ftoken\stoken|
will all be true, assuming that |*| has category~12; but if |*| has
category~10, |\if\fspace*| will be false, because \TeX\ normalizes
all newly created space tokens to \]$_{10}$, as explained in Chapter~8.) \
Since the various forms of space tokens are almost identical in
^^|\noexpand|^^|\escapechar|^^|\string|
behavior, there's no point in dwelling on the details.\footnote\dag{The
following little program is for \TeX\ exegetes who insist on
learning the whole story: Macro |\stest| decides whether or not
a given token list register begins with
a \<space token> as defined in Chapter~24. If so, the macro
decides whether the token is explicit and/or funny and/or active.
\belowdisplayskip=-\baselineskip % cancel unwanted strut at end of footnote
\begintt|global|displayindent=0pt|global|displaywidth=|hsize
\newif\ifspace \newif\iffunny \newif\ifexplicit \newif\ifactive
\def\stest#1{\funnyfalse \expandafter\s\the#1! \stest}
\def\s{\global\explicitfalse \global\activefalse \futurelet\next\ss}
\def\ss{\ifcat\noexpand\next\stoken\let\nxt\sx\else\let\nxt\ns\fi\nxt}
\def\sx{\spacetrue\ifx\next\stoken\let\nxt\sss\else\let\nxt=\ssss\fi\nxt}
\long\def\sss#1 #2\stest{\def\next{#1}%
  \ifx\next\empty \global\explicittrue \else\testactive#1\s\fi}
\long\def\ssss#1#2\stest{\funnytrue{\escapechar=\if*#1`?\else`*\fi\relax
  \if#1\string#1\uccode`#1=`~ % we assume that ~ is an active character
    \uppercase{\ifcat\noexpand#1}\noexpand~\global\activetrue
    \else\global\explicittrue\fi
  \else\testactive#1\s\fi}}
\long\def\ns#1\stest{\spacefalse}
\long\def\testactive#1#2\s{\expandafter\tact\string#1\s\tact}
\long\def\tact#1#2\tact{\def\next{#2}\ifx\next\xs\global\activetrue
  \else\ifx\next\empty \global\activetrue\fi\fi} \def\xs{\s}
\endtt}

\smallbreak
The argument to ^|\write| is expanded when a |\shipout| occurs, but
sometimes expansion isn't desired. Here's a macro (suggested by Todd
^{Allen}) that suppresses all expansion, by inserting ^|\noexpand| before
each control sequence or ^{active character}. The macro assumes that
|~|~is an active character, and that the tokens being written
do not include implicit spaces or braces. Funny spaces are changed
to ordinary ones.
\begintt
\long\def\unexpandedwrite#1#2{\def\finwrite{\write#1}%
  {\aftergroup\finwrite\aftergroup{\sanitize#2\endsanity}}}
\def\sanitize{\futurelet\next\sanswitch}
\def\sanswitch{\let\n@xt\endsanity \ifx\next\endsanity
  \else\ifcat\noexpand\next\stoken\aftergroup\space\let\n@xt=\eat
   \else\ifcat\noexpand\next\bgroup\aftergroup{\let\n@xt=\eat
    \else\ifcat\noexpand\next\egroup\aftergroup}\let\n@xt=\eat
     \else\let\n@xt=\copytok\fi\fi\fi\fi \n@xt}
\def\eat{\afterassignment\sanitize \let\next= }
\long\def\copytok#1{\ifcat\noexpand#1\relax\aftergroup\noexpand\fi
  \ifcat\noexpand#1\noexpand~\aftergroup\noexpand\fi
  \aftergroup#1\sanitize}
\def\endsanity\endsanity{}
\endtt
As before, the heavy use of\/ ^|\aftergroup| in |\unexpandedwrite| means
that parameter |#2| should not include more than about 150~tokens.

\subsection List macros. The next several macros we shall discuss can be
used to maintain lists of information in the form ^^{backslash backslash}
\begindisplay
|\\{|\<item$_1$>|}\\{|\<item$_2$>|}| $\ldots$ |\\{|\<item$_n$>|}|
\enddisplay
where each \<item> is a balanced list of tokens. A parameterless control
sequence whose replacement text has this form may be called a {\sl ^{list
macro}}. The empty list macro has $n=0$ and it is called ^|\empty|.

It's easy to add new items at either end of a list macro, and to
concatenate list macros, for example as follows: ^^|\toksdef| ^^|\long|
\begintt
\toksdef\ta=0 \toksdef\tb=2 % token list registers for temp use
\long\def\leftappenditem#1\to#2{\ta={\\{#1}}\tb=\expandafter{#2}%
  \edef#2{\the\ta\the\tb}}
\long\def\rightappenditem#1\to#2{\ta={\\{#1}}\tb=\expandafter{#2}%
  \edef#2{\the\tb\the\ta}}
\def\concatenate#1=#2&#3{\ta=\expandafter{#2}\tb=\expandafter{#3}%
  \edef#1{\the\ta\the\tb}}
\endtt
Conversely, the left item of a list can be removed and placed in a
control sequence by the |\lop| macro defined in the following curious way:
\begintt
\def\lop#1\to#2{\expandafter\lopoff#1\lopoff#1#2}
\long\def\lopoff\\#1#2\lopoff#3#4{\def#4{#1}\def#3{#2}}
\endtt
For example, if\/ |\l| expands to the list `|\\{a\b}\\{c}\\{{d}}|', the
macro invocation |\lop\l\to\z| makes |\l| expand to `|\\{c}\\{{d}}|'
and |\z|~expand to `|a\b|'. The |\lop| operation should be used only when
|\l|~is nonempty, otherwise an error will occur; to test if\/ |\l|~is
empty, one simply says `|\ifx\l\empty|'.

The programming details of the |\lop| macro indicate why individual
items have been enclosed in |{...}| ^{groups}. A simpler kind of list,
in which grouping is omitted and an extra |\\| appears at the end,
suffices for many purposes; one could define, for instance,
\begintt
\long\def\lopoff\\#1\\#2\lopoff#3#4{\def#4{#1}\def#3{\\#2}}
\endtt
and the results would be almost the same as before. In this case an empty
list macro expands to `|\\|'. However, the new |\lop| resulting from this
new |\lopoff| macro also removes a pair
of braces, if the leftmost item happens to be a group; extra braces are
included in our general scheme to prevent such anomalies.

So far the examples we've considered haven't revealed why the |\\|'s
appear in the general scheme; it appears that grouping by itself
should be enough. But in fact, the |\\| separators are enormously
useful, because we can define |\\| to be any desired one-argument macro,
and then we can {\sl execute\/} the list! For example, here's a
way to count the number of items: ^^{hash hash}
\begintt
\def\cardinality#1\to#2{#2=0 \long\def\\##1{\advance#2 by1 }#1}
\endtt
(Parameter |#2| is supposed to be the name of a count register.) \
And here's a way to take a list macro and center all its items
on individual lines within a |\vbox|:
\begintt
\def\centerlist#1{\def\\##1{\relax##1\cr}%
  \vbox{\halign{\hfil##\hfil\cr#1}}}
\endtt
A particular item can be selected by its position number from the left:
\begintt
\def\select#1\of#2\to#3{\def#3{\outofrange}%
  \long\def\\##1{\advance#1-1 \ifnum#1=0 \def#3{##1}\fi}#2}
\endtt
(Here |#1| is a count register, |#2| is a list macro, and |#3| is a
control sequence.) \ And so on; hundreds of other applications can
be imagined.\footnote\dag{The concept of a list macro is strongly related
to the concept of a list procedure in a programming language;
see {\sl Communications of the ACM\/ \bf7} (1964), 280.}

\TeX\ does all of the preceding operations efficiently, in the sense
that the running time will be proportional to the length of the list
macro involved. It's natural to ask if the rightmost item can be
removed with equal efficiency, since the final item of a list is
somewhat hard to isolate. There is apparently no way to delete the
$n$th item of an $n$-item list in order~$n$ steps, maintaining
complete generality, unless the ^|\aftergroup| trick (by which
we created a macro that expands to $n$~asterisks) is used; and the
|\aftergroup| trick is somewhat unattractive in the list application,
because the list might be quite long.\footnote\ddag{The interested reader may
enjoy constructing a macro that removes the $k$th item of an $n$-item list
macro~|\l| in $O(n\log n)$ steps, given $k$ and~|\l|,
without using |\aftergroup|.} However, if we restrict list items to
unexpandable tokens, it turns out to be possible to remove the
^^|\xyzzy| rightmost item quite efficiently:
\begintt
\def\deleterightmost#1{\edef#1{\expandafter\xyzzy#1\xyzzy}}
\long\def\xyzzy\\#1#2{\ifx#2\xyzzy\yzzyx
  \else\noexpand\\{#1}\fi\xyzzy#2}
\long\def\yzzyx#1\xyzzy\xyzzy{\fi}
\endtt
Careful study of this example shows that \TeX's ^{mouth} is capable of
doing ^{recursive} operations, given sufficiently tricky macros.

The contents of a ^|\count| register can easily be converted to decimal
and stored in a control sequence; for example, if\/ |\n| is a register,
`|\edef\csn{\the\n}|' puts its value into~|\csn|. Conversely, a value
from |\csn| can be put back into |\n| by saying simply `|\n=\csn|'.
There's usually no point in doing this transformation just to minimize
the usage of\/ |\count| registers, since \TeX\ has 256 of them; but a
decimal representation like the expansion of\/ |\csn| can be stored in a
list macro, and that might be useful in some applications. Incidentally,
there's a neat way to test if such a control-sequence-number is zero:
`^|\if||0\csn|\<true text>|\else|\<false text>|\fi|' works because
extra digits of a nonzero number will be ignored with the \<true text>.

A technique something like list macros can be used to maintain
unordered sets of control sequences. In this case it's
convenient to leave off the braces; for example,
\begintt
\def\l{\\\alpha\\\beta\\\gamma}
\endtt
defines a ``^{set macro}'' |\l| that represents the control
sequences $\{\,\hbox{|\alpha|},\hbox{|\beta|},\hbox{|\gamma|}\,\}$.
A straightforward construction tests whether a given control sequence
is in the set:
\begintt
\def\ismember#1\of#2{\resultfalse\def\given{#1}%
  \def\\##1{\def\next{##1}\ifx\next\given\resulttrue\fi}#2}
\endtt
And an efficient but not-so-straightforward construction removes all
occurrences of control sequences that are
|\ifx|-equivalent to a given control sequence:
\begintt
\def\remequivalent#1\from#2{\let\given=#1%
  \ifx#2\empty\else\edef#2{\expandafter\plugh#2\plugh}\fi}
\def\plugh\\#1#2{\ifx#1\given\else\noexpand\\\noexpand#1\fi
  \ifx#2\plugh\hgulp\fi\plugh#2}
\def\hgulp\fi\plugh\plugh{\fi}
\endtt

\subsection ^{Verbatim} listing. Plain \TeX\ includes a macro called
^|\dospecials| that is essentially a set macro, representing the set
of all characters that have a special category code. \ (The control
sequence ^|\do| plays the r\^ole of\/ |\\| in the discussion above.) \
Therefore it's easy to change all of the special characters to
category~12~(other):
\begintt
\def\uncatcodespecials{\def\do##1{\catcode`##1=12 }\dospecials}
\endtt
This works even when the set of special characters has been changed,
provided that |\dospecials| has been updated to represent the current set.

The operation |\uncatcodespecials| just defined is important, of course,
when \TeX's automatic features need to be temporarily disabled. Let's
suppose that we want to create a listing of some computer file,
reproducing the characters and the spacing exactly as they appear in the
file. To make the problem more interesting, let's also print line
numbers in front of each line, as in the listing of |story.tex| on
page~\storypage. To make the problem simpler, let's assume that the file
contains only standard ASCII printing characters: no tab marks or
form feeds or such things. Our goal is to devise a |\listing| macro
such that, e.g., `|\listing{story}|' will insert a listing of
the |story.tex| file into a manuscript, after which \TeX's normal
conventions will be restored. The listing should be in ^|\tt| type.
A macro of the following form meets the desired specifications:
\begintt
\def\listing#1{\par\begingroup\setupverbatim\input#1 \endgroup}
\endtt
Notice that the ^|\endgroup| command here will nicely ``turn off'' all
the weird things that ^|\setupverbatim| turns on. Notice also that the
commands `|\input#1 \endgroup|' will not be listed verbatim, even though
they follow |\setupverbatim|, since they entered \TeX's reading mechanism
when the |\listing| macro was expanded (i.e., before the verbatim business
was actually set up).

But what should |\setupverbatim| do? Well, it ought to include
^|\obeylines|, since this automatically inserts a ^|\par| at the end of
each line that is input; it ought to include ^|\uncatcodespecials|, so
that special characters print as themselves; and it ought to include
^|\obeyspaces|, so that each space counts. But we need to look carefully
at each of these things to see exactly what they do: \
(1)~Plain \TeX's |\obeylines| macro changes the ^|\catcode| of |^^M|
to |\active|, and then it says `|\let^^M=\par|'. Since |^^M| is placed
at the end of each line, this effectively ends each line with |\par|;
however, |\obeylines| doesn't say `|\def^^M{\par}|', so we must make
any desired changes to |\par| before invoking |\obeylines|. \
(2)~The |\uncatcodespecials| operation changes a space to category~12;
but the |\tt| font has the character `\]' in the \<space> position,
so we don't really want \]$_{12}$. \
(3)~The |\obeyspaces| macro in Appendix~B merely changes the \<space>
character to category~13; ^{active character} \]$_{13}$ has been
defined to be the same as ^|\space|, a macro that expands to \]$_{10}$.
This is usually what is desired; for example, it means that spaces in
constructions like `|\hbox to 10 pt {...}|' won't cause any trouble.
But in our application it has an undesirable effect, because it
produces spaces that are affected by the ^{space factor}. To defeat this
feature, it's necessary either to say ^|\frenchspacing| or to
redefine \]$_{13}$ to be the same as |\|\]. ^^{control space}
The latter alternative is better, because the former will discard spaces
at the beginning of each line.

The |\setupverbatim| macro should also take care of putting a line
number into the position of the paragraph indentation. We can take care
of this by introducing a counter variable and using ^|\everypar|, as follows:
\begintt
\newcount\lineno % the number of file lines listed
\def\setupverbatim{\tt \lineno=0
  \obeylines \uncatcodespecials \obeyspaces
  \everypar{\advance\lineno by1 \llap{\sevenrm\the\lineno\ \ }}}
{\obeyspaces\global\let =\ } % let active space = control space
\endtt
^^|\llap|
In theory, this seems like it ought to work; but in practice, it fails
in two ways. One rather obvious failure---at least, it becomes obvious
when the macro is tested---is that all the empty lines of the file
are omitted. The reason is that the |\par| command at the end of an empty
line doesn't start up a new paragraph, because it occurs in vertical mode.
The other failure is not as obvious, because it occurs much less often:
The |\tt| fonts contain ^{ligatures} for Spanish punctuation, so the
sequences |?||`| and~|!||`| will be printed as {\tt?`} and~{\tt!`}
respectively.  Both of these defects can be cured by inserting
\begintt
\def\par{\leavevmode\endgraf} \catcode`\`=\active
\endtt
before |\obeylines| in the |\setupverbatim| macro, and by defining |`|$_{13}$
as follows:
\begintt
{\catcode`\`=\active \gdef`{\relax\lq}}
\endtt
A similar scheme could be used to produce verbatim listings in other fonts;
but more characters would have to be made active, in order to break
ligatures and to compensate for ASCII characters that aren't present.

\def\uncatcodespecials{\def\do##1{\catcode`##1=12 }\dospecials}
\def\verbatim{\begingroup\tt\uncatcodespecials\obeyspaces\doverbatim}
\newcount\balance
{\catcode`<=1 \catcode`>=2 \catcode`\{=12 \catcode`\}=12
  \gdef\doverbatim{<\balance=1\verbatimloop>
  \gdef\verbatimloop#1<\def\next<#1\verbatimloop>%
    \if#1{\advance\balance by1
    \else\if#1}\advance\balance by-1
     \ifnum\balance=0\let\next=\endgroup\fi\fi\fi\next>>
Instead of listing a file verbatim, you might want to define a
|\verbatim| macro such that `|\verbatim{$this$|{\tt\ is }|{\it!}}|'
yields `\verbatim{$this$ is {\it!}}'. It's somewhat dangerous to change
^{category codes}, because \TeX\ stamps the category on each character
when that character is first read from a file. Thus, if\/ |\verbatim|
were defined by a construction of the form
|\long\def\verbatim#1{|\<something>|}|, argument~|#1| would already be
converted to a list of tokens when \<something> starts; |\catcode| changes
would not affect the argument.  The alternative is to change category
codes before scanning the argument to |\verbatim|:
\begintt
\def\verbatim{\begingroup\tt\uncatcodespecials
  \obeyspaces\doverbatim}
\newcount\balance
{\catcode`<=1 \catcode`>=2 \catcode`\{=12 \catcode`\}=12
  \gdef\doverbatim{<\balance=1\verbatimloop>
  \gdef\verbatimloop#1<\def\next<#1\verbatimloop>%
    \if#1{\advance\balance by1
    \else\if#1}\advance\balance by-1
     \ifnum\balance=0\let\next=\endgroup\fi\fi\fi\next>>
\endtt
This works; but it's slow, and it allows verbatim setting only of text
that has balanced braces. It would not be suitable for typesetting
the examples in a book like {\sl The \TeX book}. \ (Appendix~E contains
the verbatim macros that were actually used.) \ Note also that if this
|\verbatim{...}| macro appears in the argument to another macro
like |\centerline|, it will fail because the category codes can no longer be
changed.  The ^|\footnote| macro in Appendix~B is careful to avoid
scanning its argument prematurely; it uses ^|\bgroup| and ^|\egroup| in a
somewhat tricky way, so that category code changes are permitted inside
plain \TeX\ footnotes.

On the other hand, there is a fairly fast way to convert a token list
to an almost-verbatim transcript:
\begintt
\long\def\verbatim#1{\def\next{#1}%
  {\tt\frenchspacing\expandafter\strip\meaning\next}}
\def\strip#1>{}
\endtt
Tokens are stripped off in this construction since, for example,
^|\meaning||\next| might be `\def\next{$this$ is {\it!}}%
\hbox{\kern1pt\expandafter|\meaning\next|}'.
Notice that a space will be inserted after
the control word |\it|, but no space might actually have occurred there in
the argument to |\verbatim|; such information has been irretrievably lost.

One of the problems with verbatim mode is that it's hard to stop;
if we turn off all of \TeX's normal control capabilities, we end up
``painting ourselves into a corner'' and reaching a point of no return.
The |\listing| macro was able to solve this problem because the end of
a file brings an old token list back to life. Another solution would be
to specify a certain number of lines, after which verbatim mode should
end. Otherwise it's necessary to put some constraint on the text,
i.e., to make certain texts unprintable in verbatim mode. For example,
here's an approach that typesets everything between |\beginverbatim| and
|\endverbatim|, assuming only that the control sequence |\endverbatim|
does not need to be set:
\begintt
\def\beginverbatim{\par\begingroup\setupverbatim\doverbatim}
{\catcode`\||=0 \catcode`\\=12 % || is temporary escape character
  ||obeylines||gdef||doverbatim^^M#1\endverbatim{#1||endgroup}}
\endtt
This construction assumes that |\beginverbatim| appears at the end of a
line in the manuscript file.  Argument |#1| will be read entirely into
\TeX's memory before anything happens, so the total amount of verbatim
material had better not be too voluminous. Incidentally, it isn't
necessary to say that this macro is ^|\long|, because the |\par|'s
inserted by ^|\obeylines| are really |^^M|'s.

Another approach is to keep one character untouchable. For example,
it's possible to define things so that `|\verbatim|\<char>\<text>\<char>'
will typeset the \<text> verbatim, where the \<text> is not supposed
to contain any occurrences of the repeated delimiter \<char>:
\begintt
\def\verbatim{\begingroup\setupverbatim\doverbatim}
\def\doverbatim#1{\def\next##1#1{##1\endgroup}\next}
\endtt

\subsection Selective loading of macros. Some interesting problems arise
when a computer system acquires a large library of macro files. For
example, suppose that a file |macs.tex| contains the lines
\begintt
\let\italcorr=\/
\def\/{\unskip\italcorr}
\endtt
because somebody thought it would be nice to allow an optional space
before \TeX's primitive ^|\/| command. That's fine, except if |macs.tex|
is input twice; for example, two other macro files might both say
^|\input|~|macs|. When those lines are processed the second time,
|\italcorr| will be |\let| equal to a macro that expands to
`|\unskip\italcorr|',
and you can guess what will happen: \TeX\ will get into an infinite loop,
stoppable only by ^{interrupt}ing the program manually.
^^{recursion, infinite}

Fortunately there's an easy way to prevent this problem, by placing
a suitable interlock near the beginning of every macro file that
might introduce such anomalies:
\begintt
\ifx\macsisloaded\relax\endinput\else\let\macsisloaded=\relax\fi
\endtt
Then |\macsisloaded| will be undefined at the time of the first~|\ifx|,
but the file will not be read twice. A different control sequence
should, of course, be used for each file.

Another difficulty with large sets of macros is that they take up space.
It would be nice to preload every macro that every \TeX\ user has ever
dreamed up; but there might not be enough room,
because \TeX's memory capacity is finite. You might find it necessary
to hold back and to load only the macros that are really needed.

How much memory ^{space} ^^{efficiency} does a macro require?
Well, there are four kinds of memory involved: token memory,
name memory, string memory, and character memory. \ (If any of these
becomes too full, it will be necessary to increase what \TeX\ calls the
macro memory size, the hash size, the number of strings, and/or the
pool size, respectively; see Chapter~27.) \ The token memory is
most important; a macro takes one cell of token memory for each token
in its definition, including the `|{|' and the `|}|'. For example,
the comparatively short definition
\begintt
\def\example#1\two{\four}
\endtt
takes five tokens: |#1|, \cstok{two}, |{|$_1$, \cstok{four}, and |}|$_2$.
Each control sequence also takes up one cell of name memory, one cell
of string memory, and as many cells of character memory as there are
characters in the name (seven in the case of\/ |\example|). Character
memory is comparatively cheap; four characters, or in some cases five,
will fit in the same number of bits as a single cell of token memory,
inside the machine. Therefore you don't save much by choosing short
macro names.

\TeX\ will tell you how close you come to exceeding its current
memory capacity if you say ^|\tracingstats||=1|. For example, one of the
runs that the author made while testing galley proofs of this appendix
reported the following statistics:
\begintt
Here is how much of TeX's memory you used:
 209 strings out of 1685
 1659 string characters out of 17636
 27618 words of memory out of 52821
 1172 multiletter control sequences out of 2500
\endtt
Consequently there was plenty of room for more macros: $52821-27618=
25203$ unused cells of main memory, $2500-1172=1328$ of name memory,
$1685-209=1476$ of string memory, and $17636-1659=15977$ of character memory.
But a fairly large \TeX\ was being used, and only the macros of
Appendices B and~E were loaded; in other circumstances it might have
been necessary to conserve space.

One obvious way to keep from loading too many macros is to keep
the macro files short and to |\input| only the ones that you need.
But short files can be a nuisance; sometimes there's a better way.
For example, let's suppose that a file contains five optional classes
of macros called |A|, |B|, |C|, |D|,~|E|, and that a typical user
will probably want only at most two or three of these five; let's
design a |\load| macro so that, for example,
`|\load{macs}{AC}|'
will load file |macs.tex| including options |A| and |C| but not options
|B|, |D|, or |E|. The following |\load| macro converts its second argument
into a set~macro called |\options|:
\begintt
\def\load#1#2{\let\options=\empty \addoptions#2\end \input#1 }
\def\addoptions#1{\ifx#1\end \let\next=\relax
  \else\let\\=\relax\edef\options{\options\\#1}%
   \let\next=\addoptions \fi \next}
\endtt
Inside the file |macs.tex|, a portion of code that should be loaded only
under option~|B|, say, can be enclosed by `|\ifoption B ... \fi|', where
|\ifoption| is defined thus:
\begintt
\def\ifoption#1{\def\\##1{\if##1#1\resulttrue\fi}%
  \resultfalse \options \ifresult}
\endtt
(This is a simple application of ideas presented earlier in this appendix.)

However, the |\ifoption...\fi| scheme isn't very robust, because it requires
all of the macros in the optional part to be well nested with
respect to |\if...| and |\fi|; a macro like |\ifoption| itself couldn't
easily be defined in such a place! There's a better scheme that also
runs faster, based on ^{category code} changes. This idea (due to
Max ^{D\'\i az}) requires that the leftmost nonblank character on each
line be either `|\|' or~`|{|'; it's usually easy to arrange this.
Furthermore, one other symbol, say |~|, is reserved. Then the text
material that is to be loaded only under option~|B| is preceded by
the line `|\beginoption|~|B|' and followed by a line that says
`|~endoptionalcode|'. The ^|\catcode| for~|~| is set to~14 (comment
character), hence the |~endoptionalcode| line will have no effect if
code is not being skipped. The |\beginoption| macro works like this:
\begintt
\def\beginoption#1{\ifoption#1\else\begingroup\swapcategories\fi}
\def\swapcategories{\catcode`\\=14 \catcode`\{=14 \catcode`\~=0 }
\let\endoptionalcode=\endgroup
\catcode`\~=14
\endtt
Once the categories have been swapped, all lines will be skipped at high speed
until the control sequence |~endoptionalcode| is encountered; then
everything will be restored to its former state. Under this scheme,
material that should be loaded only under both options |B| and~|D|
can be prefaced by both `|\beginoption|~|B|' and `|\beginoption|~|D|';
material that should be loaded under either option~|B| or option~|D|
(or both) can be prefaced by
\begintt
\beginoption B
~oroption D
\endtt
if we define |\oroption#1| to be an abbreviation for
`|\ifoption#1\endgroup\fi|'.

Another kind of selective loading is sometimes appropriate, based on
whether or not a particular control sequence is defined. In this
scheme, if the control sequence is undefined, it should remain
undefined and it should take up no space whatever in \TeX's memory.
There's a slick way to do this, namely to say
\begintt
\ifx\cs\undefined ... \fi
\endtt
(assuming that ^|\undefined| has never been defined). \TeX\ does not
put undefined control sequences into its internal tables if they follow
^|\ifx| or if they are encountered while skipping ^{conditional text}.
You can use this idea, for example, to prepare a bibliography for a paper,
by reading a suitably arranged bibliography file; only the entries
that correspond to defined control sequences will be loaded.

\subsection Brace hacks. Several of \TeX's operations depend on
^{grouping}, and you'll want to know exactly what this means if
you try to do certain tricky things. For example, plain \TeX's
control sequences ^|\bgroup| and ^|\egroup| are ``^{implicit braces}''
because they have been defined by
\begintt
\let\bgroup={  \let\egroup=}
\endtt
This means that you can include them in the replacement texts of
definitions without worrying about how they nest; for example, the
macros
\begintt
\def\beginbox{\setbox0=\hbox\bgroup}
\def\endbox{\egroup\copy0 }
\endtt
allow you to make a box between |\beginbox| and |\endbox|; the behavior
is almost the same as
\begintt
\def\beginbox#1\endbox{\setbox0=\hbox{#1}\copy0 }
\endtt
but different in three important ways: (1)~The first alternative allows
category codes to change inside the box. (2)~The first alternative
is faster, because it doesn't need to scan the box contents both as an
argument and as a sequence of actual commands. (3)~The first alternative
takes less memory space, because no argument needs to be stored.
Thus, the first alternative is usually superior.

For the purposes of this discussion we shall assume that only `|{|'
has category~1 and that only `|}|' has category~2, although any
characters can actually be used as group delimiters. ^^{braces}
Group nesting is crucial during two of \TeX's main activities:
(a)~when \TeX\ is scanning a ^\<balanced text>, e.g., when \TeX\ is
forming the replacement text of a macro, a parameter, or a token list
variable; (b)~when \TeX\ must determine whether the token
|&|~or ^|\span| ^^{ampersand}
or ^|\cr| or~^|\crcr| is the end of an entry within an ^{alignment}.

\TeX's mouth has two internal counting mechanisms to deal with nesting:
The ``master counter'' goes up by~1 for each |{|$_1$ scanned by
\TeX, and down by~1 for each~|}|$_2$; the ``balance counter'' is
similar, but it is affected only by explicit |{|$_1$ and |}|$_2$ tokens
that are actually contributed to a token list that is being formed.
The master counter decreases by~1 when \TeX\ evaluates the
^{alphabetic constant} |`{|, and it increases by~1 when \TeX\ evaluates
|`}|, hence the net change is zero when such constants are evaluated.
As a consequence of these rules, certain constructions produce
the following effects:
$$\halign{\indent\hfil#\hfil&\quad\hfil#\hfil&
  \quad\hfil#\hfil&\qquad\hfil#\hfil&\quad\hfil#\hfil\cr
&\it Master counter change\span\omit&\it Balance counter change\span\omit\cr
\it Input&\it expanded&\it unexpanded&\it expanded&\it unexpanded\cr
|{|&1&1&1&1\cr
|\bgroup|&0&0&0&0\cr
|\iffalse{\fi|&1&1&0&1\cr
|\ifnum0=`{\fi|&0&1&0&1\cr}$$
The last two cases produce no begin-group tokens when expanded, but they
do affect the master counter as shown. Thus, for example,
\begintt
\def\eegroup{\ifnum0=`{\fi}}
\endtt
makes |\eegroup| behave rather like |\egroup|, but the expansion of\/
|\eegroup| also decreases the master counter.

Alignment processing uses only the master counter, not the balance counter.
An alignment entry ends with the first |&|~or |\span| or |\cr| or~|\crcr|
that appears when the master counter has the value that was present
in the counter at the beginning of the entry. Thus, for example,
the curious construction
\begintt
\halign{\show\par#\relax\cr
  \global\let\par=\cr
  {\global\let\par=\cr}\cr
  \par}
\endtt
causes \TeX\ to perform three |\show| instructions, in which the
respective values of\/ |\par| shown are |\par|, |\relax|, and~|\cr|.
Similarly, each template in the preamble to an alignment ends with
the first |&| or |\cr| or |\crcr| that appears at the master counter level
that was in effect at the beginning of the entry; hence |&|~and |\cr| and
|\crcr| tokens can appear within a template of an alignment, if they
are hidden by braces (e.g., if they appear in a definition).

These facts allow us to draw two somewhat surprising conclusions: (1)~If
an alignment entry has the form `$\,\alpha\,$^|\iffalse||{\fi|$\,\beta\,$%
|\iffalse}\fi|$\,\gamma\,$', it's possible for $\beta$ to include
|&|~and~|\cr| tokens that aren't local to a group.\footnote*{The
token list $\alpha$ should not be empty, however, because \TeX\ expands
the first token of an alignment entry before looking at the template,
in order to see if the entry begins with |\noalign| or |\omit|. The master
counter value that is considered to be present at the beginning of an
entry is the value in the counter just after the ``$u$~part'' of the
template has been entirely read.} (2)~The construction
\begintt
{\span\iffalse}\fi
\endtt
appearing in a preamble contributes `|{|' to the template without
any net change to the master counter; thus, it's very much like
|\bgroup|, except that it produces |{|$_1$ explicitly.
If you understand (1) and~(2), you'll agree that the present appendix
deserves its name.

\def\pmb#1{\setbox0=\hbox{#1}%
  \kern-.025em\copy0\kern-\wd0\kern.05em\copy0
  \kern-\wd0\kern-.025em\raise.0433em\box0 }
\subsection ^{Box maneuvers}. Let's turn now from syntax to semantics,
i.e., from \TeX's mouth to its gastro-intestinal tract. Sometimes an odd
symbol is needed in boldface type, but it's available only in a normal
weight. In such cases you can sometimes get by with ``^{poor man's bold},''
obtained by overprinting the normal weight symbol with slight offsets.
The following macro typesets its argument three times in three
slightly different places, equidistant from each other; but the result
takes up just as much space as if\/ |\pmb| had been simply |\hbox|:
\begintt
\def\pmb#1{\setbox0=\hbox{#1}%
  \kern-.025em\copy0\kern-\wd0
  \kern.05em\copy0\kern-\wd0
  \kern-.025em\raise.0433em\box0 }
\endtt
For example, `|\pmb{$\infty$}|' yields `\pmb{$\infty$}'. The results
are somewhat \pmb{fuzzy}, and they certainly are no match for the
real thing if it's available; but poor man's bold is better than nothing,
and once in a~while you can get away with it.

When you put something into a box register, you don't need to put
the contents of that register into your document. Thus, you can write
macros that do experiments behind the scenes, trying different
possibilities before making a commitment to a particular decision.
For example, suppose you are typesetting a text in two languages, and
you would like to choose the column widths so that the same number of
lines is obtained in both cases. For example, the following texts
balance perfectly when the first column is $157.1875\pt$ wide and the
second column is $166.8125\pt$ wide; but the second column would be
one line longer than the first if they were both $162\pt$ wide:
^^{Alice} ^^{Bill}
\medskip
\hrule
\smallskip
\begingroup
\eightpoint
\tolerance=9999
\hbadness=2300
\finalhyphendemerits=3000000
\doublehyphendemerits=1000000
\parskip=1pt
\parindent=1.5em
\frenchspacing

\def\firstcol{ % from Surreal Numbers, pp. 46--47
\item{A.}\strut The creative part is really more interesting than the
deductive part. Instead of concentrating just on finding good
answers to questions, it's more important to learn how to find
good questions!

\item{B.} You've got something there. I wish our teachers would
give us problems like, ``Find something interesting about $x$,''
instead of ``Prove $x$.''

\item{A.}Exactly. But teachers are so conservative, they'd be
afraid of scaring off the ``grind'' type of students who
obediently and mechanically do all the homework. Besides, they
wouldn't like the extra work of grading the answers to
nondirected questions.

\item{}The traditional way is to put off all creative aspects
until the last part of graduate school. For seventeen or more
years, a student is taught exams\-man\-ship, then suddenly after
passing enough exams in graduate school he's told to do
something original.
}
\def\\#1{\leavevmode\raise.5pt\hbox{$\scriptscriptstyle
    \ifx#1`\langle\!\langle\else\rangle\!\rangle\fi$}% Spanish quote marks
  \ifx#1`\nobreak\hskip0pt \fi} % allow hyphenation
%\showhyphens{\\`Demuestre\\' \\`apisonadora\\' \\`Encuentren\\'}
\def\secondcol{ % from N\'umeros Surreales, pp. 39--40
\item{A.} \strut La parte creativa es mucho mejor que la deductiva.
En vez de concentrarse en buscar buenas respuestas a ciertas
cuestiones es m\'as importante aprender a proponerse buenas
preguntas.

\item{B.} Me parece una buena ocurrencia. Me gustar\'\i a
que los profesores propusieran problemas del estilo de
\\`Encuentren algo interesante sobre $x$\\' en vez de
\\`Demuestre que $x\ldots\,$\\'.

\item{A.} Exactamente. Pero los profesores son tan conservadores
que temer\'\i an espantar al tipo de estudiante \\`apisonadora\\'
que hace lo que le proponen para casa, obe\-dien\-te\-mente y de forma
mec\'anica. Adem\'as, no creo que les gustase el trabajo adicional
de calificar respuestas a preguntas abiertas.

\item{}La forma tradicional es dejar la parte creativa para los cursos
altos. Durante diecisiete a\~nos o m\'as se ense\~na al es\-tu\-diante a
aprobar, luego de golpe, cerca de la graduaci\'on, se le pide que haga
algo original.
}
\newif\iffail \newdimen\doublewidth \newdimen\delheight
\newcount\n
\newdimen\trialwidth \newdimen\lowwidth \newdimen\highwidth
\wlog{Beginning the binary search for balanced columns:}
\def\balancetwocols{\lowwidth=10em % the lowest feasible \trialwidth
  \highwidth=\doublewidth \advance\highwidth-10em % the highest ditto
  {\n=1 \hbadness=10000 \hfuzz=\maxdimen
    \loop \maketrial
    \wlog{trial \the\n, \the\trialwidth: (\the\ht0,\the\ht2)}
    \testfailure \iffail \preparenewtrial \repeat}
  \maketrial} % now under/overfull boxes will be shown
\def\maketrial{%
  \trialwidth=.5\lowwidth \advance\trialwidth by.5\highwidth
  \setbox0=\vbox{\hsize=\trialwidth \firstcol}
  \setbox2=\vbox{\hsize=\doublewidth
    \advance\hsize-\trialwidth \secondcol}}
\def\testfailure{\dimen0=\ht0 \advance\dimen0-\ht2
  \ifnum\dimen0<0 \dimen0=-\dimen0 \fi
  \ifdim\dimen0>\delheight \ifnum\n=10 \failfalse\else\failtrue\fi
  \else\failfalse\fi}
\def\preparenewtrial{\ifdim\ht0>\ht2 \global\lowwidth=\trialwidth
  \else\global\highwidth=\trialwidth\fi \advance\n by1 }
\doublewidth=27pc \delheight=0pt
\balancetwocols
\line{\box0\hfil\box2}
\endgroup
\smallskip
\hrule
\medskip
Some implementations of \TeX\ display the output as you are running,
so that you can choose column widths interactively until a suitable
balance is obtained. It's fun to play with such systems, but it's also
possible to ask \TeX\ to compute the column widths automatically. The
following code tries up to ten times to find a solution in which the
natural heights of the two columns are different by less than
a given value, |\delheight|. The macros |\firstcol| and
|\secondcol| are supposed to generate the columns, and the sum
of column widths is supposed to be |\doublewidth|.
\beginlines
|\newdimen\doublewidth \newdimen\delheight \newif\iffail \newcount\n|
|\newdimen\trialwidth \newdimen\lowwidth \newdimen\highwidth|
|\def\balancetwocols{\lowwidth=10em % lower bound on \trialwidth|
|  \highwidth=\doublewidth \advance\highwidth-10em % upper bound|
|  {\n=1 \hbadness=10000 \hfuzz=\maxdimen % disable warnings|
|    \loop \maketrial \testfailure \iffail \preparenewtrial \repeat}|
|  \maketrial} % now under/overfull boxes will be shown|
|\def\maketrial{\trialwidth=.5\lowwidth \advance\trialwidth by.5\highwidth|
|  \setbox0=\vbox{\hsize=\trialwidth \firstcol}|
|  \setbox2=\vbox{\hsize=\doublewidth\advance\hsize-\trialwidth\secondcol}}|
|\def\testfailure{\dimen0=\ht0 \advance\dimen0-\ht2|
|  \ifnum\dimen0<0 \dimen0=-\dimen0 \fi|
|  \ifdim\dimen0>\delheight \ifnum\n=10 \failfalse\else\failtrue\fi|
|  \else\failfalse\fi}|
|\def\preparenewtrial{\ifdim\ht0>\ht2 \global\lowwidth=\trialwidth|
|  \else\global\highwidth=\trialwidth\fi \advance\n by1 }|
\endlines
Neither column will be less than 10~ems wide.  This code does a ``^{binary
search},'' assuming that a column will not increase in height when it is
made wider. If no solution is found in 10~trials, there probably is no way
to obtain the desired balance, because a tiny increase in the width of the
taller column will make it shorter than the other one.  The values of\/
^|\hbadness| and ^|\hfuzz| are made infinite during the trial settings,
because warning messages that relate to unused boxes are irrelevant; after
a solution is found, it is computed again, so that any relevant warnings
will be issued.

When a box has been put into a box register, you can change its
height, width, or depth by assigning a new value to the |\ht|, |\wd|,
or |\dp|. Such assignments don't change anything inside the box;
in particular, they don't affect the setting of the glue.

But changes to a box's dimensions can be confusing if you don't understand
exactly how \TeX\ deals with boxes in lists. The rules are stated in
Chapter~12, but it may be helpful to restate them here in a different way.
Given a box and the location of its reference point, \TeX\ assigns
locations to interior boxes as follows:  \ (1)~If the box is an hbox,
\TeX\ starts at the reference point and walks through the horizontal list
inside. When the list contains a box, \TeX\ puts the reference point of
the enclosed box at the current position, and moves right by the width of
that box.  When the list contains glue or kerning, etc., \TeX\ moves right
by the appropriate amount. \ (2)~If the box is a vbox, \TeX\ starts at the
upper left corner (i.e., \TeX\ first moves up from the reference point, by
the height of the box) and walks through the vertical list inside. When
the list contains a box, \TeX\ puts the upper left corner of that
box at the current position; i.e., \TeX\ moves down by the height of that
box, then puts the box's reference point at the current position, then
moves down by the depth of the box. When the list contains glue or
kerning, etc., \TeX\ moves down by the appropriate amount.

As a consequence of these rules, we can work out what happens when
the dimensions of a box are changed. Let |\delta| be a \<dimen>
register, and let |\h| and |\hh| specify horizontal lists that
don't depend on |\box0|. Consider the following macro:
\begintt
\newdimen\temp \newdimen\delta
\def\twohboxes#1{\setbox1=\hbox{\h \copy0 \hh}
  \temp=#10 \advance\temp by \delta #10=\temp
  \setbox2=\hbox{\h \copy0 \hh}}
\endtt
For example, |\twohboxes\wd| makes two hboxes, |\box1| and |\box2|,
that are identical except that the width of\/ |\box0| has been
increased by $\delta$ in |\box2|. What difference does this make?
There are several cases, depending on whether |#1| is |\wd|, |\ht|,
or |\dp|, and depending on whether |\box0| is an hbox or a vbox.
\ {\it Case~1}, |\twohboxes\wd|: The material from |\hh| is
moved right by $\delta$ in |\box2|, compared to its position
in |\box1|. Also |\wd2| is $\delta$ more than~|\wd1|.
\ {\it Case~2}, |\twohboxes\ht|: If\/ |\box0| is an hbox, everything
remains in the same position; but if\/ |\box0| is a vbox, everything
in |\copy0| moves up by $\delta$. Also |\ht2| may differ from |\ht1|.
\ {\it Case~3}, |\twohboxes\dp|: Everything remains in the
same position, but |\dp2| may differ from |\dp1|.

Similarly, we can work out the changes when box dimensions are
changed for boxes within vertical lists. In this case we shall
ignore the influence of interline glue by defining |\twovboxes| as follows:
\begintt
\def\twovboxes#1{
  \setbox1=\vbox{\v\nointerlineskip\copy0\nointerlineskip\vv}
  \temp=#10 \advance\temp by \delta #10=\temp
  \setbox2=\vbox{\v\nointerlineskip\copy0\nointerlineskip\vv}}
\endtt
What is the difference between |\box1| and |\box2| now?
\ {\it Case~1}, |\twovboxes\wd|: Everything remains in the same
position, but |\wd2| may differ from |\wd1|.
\ {\it Case~2}, |\twovboxes\ht|: If\/ |\box0| is an hbox, everything
in |\v| moves up by~$\delta$ in |\box2|, compared to
the corresponding positions in |\box1|, if we make the reference
points of the two boxes identical; but if\/ |\box0| is a vbox,
everything in it moves up by~$\delta$, together with the material
in |\v|. Also, |\ht2| is $\delta$~more than |\ht1|.
\ {\it Case~3}, |\twovboxes\dp|:
If\/ |\vv| is empty, |\dp2| is $\delta$~more than~|\dp1|, and
nothing else changes. Otherwise everything in |\v| and in |\copy0|
moves up by~$\delta$, and |\ht2| is $\delta$ more than~|\ht1|.
%\ (If you don't believe it, try it.)

\newdimen\unit
\def\point#1 #2 {\rlap{\kern#1\unit
    \raise#2\unit\hbox{$
      \scriptstyle\bullet\;(#1,#2)$}}}
\unit=\baselineskip
\setbox0=\vtop{\hrule
  \hbox{\vrule height10\unit depth9.4\unit \kern2\unit
    \hbox{\unit=\baselineskip
      \point 0 0 % Alioth (Epsilon Urs\ae\ Majoris), magnitude 1.79
      \point 0 8 % Dubhe (Alpha Urs\ae\ Majoris),    magnitude 1.81
      \point 0 -8 % Alkaid(Eta Urs\ae\ Majoris),     magnitude 1.87
      \point -1 -2.5 % Mizar (Zeta Urs\ae\ Majoris), magnitude 2.26
      \point 4 7 % Merak (Beta Urs\ae\ Majoris),     magnitude 2.37
      \point 4 2 % Phekda (Gamma Urs\ae\ Majoris),   magnitude 2.44
      \point 1 1.5 % Megrez (Delta Urs\ae\ Majoris), magnitude 3.30
      }%        Sources: Atlas of the Universe; Astronomy Data Book
    \kern7\unit \vrule}\hrule}
\dp0=0pt
\dimen@=240pt
\parshape=21
\z@\hsize \z@\hsize
\z@\dimen@ \z@\dimen@ \z@\dimen@ \z@\dimen@ \z@\dimen@ \z@\dimen@
\z@\dimen@ \z@\dimen@ \z@\dimen@ \z@\dimen@ \z@\dimen@ \z@\dimen@
\z@\dimen@ \z@\dimen@ \z@\dimen@ \z@\dimen@ \z@\dimen@ \z@\dimen@
\z@\hsize
\TeX\ is designed to put boxes together either horizontally
or vertically, not diagonally. But that's not a serious
limitation, because the use of negative spacing makes it
possible to put things anywhere on a page. ^^{Ursa Major}
^^{points with arbitrary coordinates} ^^{coordinates}
\vadjust{\kern-\ht0\kern-2pt\rightline{\box0}\kern2pt}%
For example, the seven points in the diagram at the right of this
paragraph were typeset by saying simply
\begintt|hsize=|displaywidth
\hbox{\unit=\baselineskip
  \point 0 0
  \point 0 8
  \point 0 -8
  \point -1 -2.5
  \point 4 7
  \point 4 2
  \point 1 1.5
  }
\endtt
{\count0=\prevgraf \advance\count0 by7 \prevgraf=\count0}%
The |\point| macro makes a box of width zero; hence the individual |\point|
specifications can be given in any order, and there's no restriction on
the coordinates:
\begintt|hsize=|displaywidth
\newdimen\unit
\def\point#1 #2 {\rlap{\kern#1\unit
    \raise#2\unit\hbox{$
      \scriptstyle\bullet\;(#1,#2)$}}}
\endtt
^^|\kern|{\count0=\prevgraf \advance\count0 by3 \prevgraf=\count0}%
If the |\point| specifications are not enclosed in an ^|\hbox|---i.e., if
they occur in vertical mode---a similar construction can be used. In
this case |\point| should create a box whose height and depth are zero:
\begintt
\def\point#1 #2 {\vbox to0pt{\kern-#2\unit
    \hbox{\kern#1\unit$\scriptstyle\bullet\;(#1,#2)$}\vss}
  \nointerlineskip}
\endtt
(The ^|\nointerlineskip| is necessary to prevent interline glue from
messing things up.)

\begingroup
\let\qc=\manual
\catcode`\ =9 \endlinechar=-1 % ignore all spaces (temporarily)
\newcount\dir \newdimen\y \newdimen\w
\newif\ifvisible \let\B=\visibletrue \let\W=\visiblefalse
\newbox\NE \newbox\NW \newbox\SE \newbox\SW \newbox\NS \newbox\EW
\setbox\SW=\hbox{\qc a} \setbox\NW=\hbox{\qc b}
\setbox\NE=\hbox{\qc c} \setbox\SE=\hbox{\qc d}
\w=\wd\SW \dimen0=\fontdimen8\qc
\setbox\EW=\hbox{\kern-\dp\SW \vrule height\dimen0 width\wd\SW} \wd\EW=\w
\setbox\NS=\hbox{\vrule height\ht\SW depth\dp\SW width\dimen0}  \wd\NS=\w
\def\L{\ifcase\dir \dy+\NW \or\dx-\SW \or\dy-\SE \or\dx+\NE\dd-4\fi \dd+1}
\def\S{\ifcase\dir \dx+\EW \or \dy+\NS \or \dx-\EW \or \dy-\NS \fi}
\def\R{\ifcase\dir \dy-\SW\dd+4 \or\dx+\SE \or\dy+\NE \or\dx-\NW\fi \dd-1}
\def\T{\ifcase\dir\kern-\w\dd+2\or\ey-\dd+2\or\kern\w\dd-2\or\ey+\dd-2\fi}
\edef\dd#1#2{\global\advance\dir#1#2\space}
\def\dx#1#2{\ifvisible\raise\y\copy#2 \if#1-\kern-2\w\fi\else\kern#1\w\fi}
\def\dy#1#2{\ifvisible\raise\y\copy#2 \kern-\w \fi \global\advance\y#1\w}
\def\ey#1{\global\advance\y#1\w}
\def\path#1{\hbox{\B \dir=0 \y=0pt #1}}
\catcode`\ =10 \endlinechar='15 % resume normal spacing conventions

If you enjoy fooling around making pictures, instead of typesetting
ordinary text, \TeX\ will be a source of endless frustration/amusement
for you, because almost anything is possible if you have suitable fonts.
For example, suppose you have a font |\qc| that contains four ^{quarter
circles}:
\begindisplay
\def\\#1{\kern\dp#1\copy#1\kern-\dp#1}
\tt a = \\\SW \qquad b = \\\NW \qquad c = \\\NE \qquad d = \\\SE
\enddisplay
Each of these characters has the same height, the same width, and the same
depth; the width and the height-plus-depth are equal to the diameter of
the corresponding full circle. Furthermore, the reference point of each
character is in a somewhat peculiar place: Each quarter arc has a
horizontal endpoint such that the lower edge of the curve is at the
baseline, and a vertical endpoint such that the left edge is directly
above or below the reference point. This convention makes it possible to
guarantee perfect alignment between these characters and rules that meet
them at the endpoints; the thickness of such rules should be
^|\fontdimen||8\qc|.

Given those characters, it's possible to devise macros |\path|, |\L|,
^^{turtle commands}
|\R|, |\S|, and~|\T| such that ^|\path||{|$\langle$any string of\/ |\L|'s, |\R|'s,
|\S|'s, and |\T|'s$\rangle$|}| produces a path that starts traveling East,
but it turns left for each~|\L|, right for each~|\R|, goes straight
for each~|\S|, and turns backward for each~|\T|. Thus, for example,
%|\path{\S\T\R\R\R\L\S\T\R\R\R\L}| yields `\kern\dp\SW
%\smash{\path{\S\T\R\R\R\L\S\T\R\R\R\L}}\kern-\dp\SW',
|\path{\L\T\S\T\R\L\T\S\T\R}| yields `\kern\dp\SW
\smash{\path{\L\T\S\T\R\L\T\S\T\R}}\kern-\dp\SW', and you
can also get the following effects:
\def\sm#1{\smash{$\vcenter{#1}$}}%
\begindisplay
\noalign{\kern5pt}
|\path{\L\R\S\R\S\R\S\S\R\R}|\kern30pt
  \sm{\path{\L\R\S\R\S\R\S\S\R\R}}\cr
\noalign{\kern16pt}
|\path{\R\R\R\R\T\S\S\L\L\L\L\L\S\S}|\kern47pt
  \sm{\path{\R\R\R\R\T\S\S\L\L\L\L\L\S\S}}\cr
\noalign{\kern16pt}
|\def\X{\L\T\L\L\T\L\L\T} \path{\X\X\X\X}|\kern20pt
  \sm{\def\X{\L\T\L\L\T\L\L\T} \path{\X\X\X\X}}\cr
\noalign{\kern10pt}
%|\path{\S\S\S\S\L\R\R\L\S\R\R\L\R\R\L\S\S\S\L\R\R\L\R\R}|\kern20pt
%  \sm{\path{\S\S\S\S\L\R\R\L\S\R\R\L\R\R\L\S\S\S\L\R\R\L\R\R}}\cr
\enddisplay
Furthermore, there are operations |\B| and |\W| that make the path
black (visible) and white (invisible), respectively:
\begindisplay \openup1pt
\noalign{\kern14pt}
|\path{\R\R\S|\cr
|  \W\S\S\S\R\R|\cr
|  \B\R\R\S\R\S\R\S\S\S\R\S\S\S\S\S\R\S\R|
 \kern100pt\raise4pt\hbox{\sm{\path{\R\R\S
  \W\S\S\S\R\R
  \B\R\R\S\R\S\R\S\S\S\R\S\S\S\S\S\R\S\R
  \W\R\R\R\S\L\S
  \B\L\S\S\S\S}}}\cr
|  \W\R\R\R\S\L\S|\cr
|  \B\L\S\S\S\S}|\cr
\noalign{\kern15pt}
\enddisplay
(It may be necessary to put kerns before and after the path, since the
box produced by |\path| may not be as wide as the actual path itself.)

The |\path| macros work differently from |\point|, since the boxes need
not have zero width in this application:
\beginlines
^|\catcode||`\ =9 |^|\endlinechar||=-1 % ignore all spaces (temporarily)|
|\newcount\dir \newdimen\y \newdimen\w|
|\newif\ifvisible \let\B=\visibletrue \let\W=\visiblefalse|
|\newbox\NE \newbox\NW \newbox\SE \newbox\SW \newbox\NS \newbox\EW|
|\setbox\SW=\hbox{\qc a} \setbox\NW=\hbox{\qc b}|
|\setbox\NE=\hbox{\qc c} \setbox\SE=\hbox{\qc d}|
|\w=\wd\SW \dimen0=\fontdimen8\qc|
|\setbox\EW=\hbox{\kern-\dp\SW \vrule height\dimen0 width\wd\SW} \wd\EW=\w|
|\setbox\NS=\hbox{\vrule height\ht\SW depth\dp\SW width\dimen0}  \wd\NS=\w|
|\def\L{\ifcase|%
  |\dir \dy+\NW \or\dx-\SW \or\dy-\SE \or\dx+\NE\dd-4\fi \dd+1}|\kern-2pt
|\def\S{\ifcase\dir \dx+\EW \or \dy+\NS \or \dx-\EW \or \dy-\NS \fi}|
|\def\R{\ifcase|%
  |\dir \dy-\SW\dd+4 \or\dx+\SE \or\dy+\NE \or\dx-\NW\fi \dd-1}|\kern-2pt
|\def\T{|^|\ifcase|%
  |\dir\kern-\w\dd+2\or\ey-\dd+2\or\kern\w\dd-2\or\ey+\dd-2\fi}|\kern-2pt
\goodbreak
|\edef\dd#1#2{\global\advance\dir#1#2\space}|
|\def\dx#1#2|%
  |{\ifvisible\raise\y\copy#2 \if#1-\kern-2\w\fi\else\kern#1\w\fi}|\kern-2pt
|\def\dy#1#2{\ifvisible\raise\y\copy#2 \kern-\w \fi \global\advance\y#1\w}|
|\def\ey#1{\global\advance\y#1\w}|
|\def\path#1{\hbox{\B \dir=0 \y=0pt #1}}|
|\catcode`\ =10 |^|\endlinechar||=`\^^M % resume normal spacing conventions|
\smallskip
|\newcount\n % the current order in the \dragon and \nogard macros|
|\def\dragon{\ifnum\n>0{\advance\n-1 \dragon\L\nogard}\fi}|
|\def\nogard{\ifnum\n>0{\advance\n-1 \dragon\R\nogard}\fi}|
\endlines
(The last three lines are not part of the |\path| macros, but they can
be used as an interesting test case. To get the famous ``^{dragon curve}''
of order~9, all you have to say is `|\path{\dir=3 \n=9 \dragon}|'.)^^{recursion}
\endgroup

\smallbreak
Let's turn now to another box-oriented problem. The ^|\listing| macro
discussed earlier in this appendix was restricted to listing files
that contain only visible ASCII characters. Sometimes it's desirable to
deal with ASCII ^\<tab> marks too, where a \<tab> is equivalent to
1~or~2~or $\cdots$ or~8 spaces (whatever is necessary to make the
current line length a multiple of~8). How can this be done?

We shall assume that files can contain a special symbol that \TeX\ will
input as character number~9, the ASCII \<tab> code; some implementations
can't actually do this. If a file contains the three symbols |^^I|, plain
\TeX\ will normally input them as a single character, number~9; but in a
verbatim listing of the file we naturally want such symbols to print as
themselves, i.e., as |^^I|.

The following construction redefines |\setupverbatim| so that the previous
|\listing| macro will work with \<tab> characters. The idea is to keep the
line-so-far in an hbox, which can be ``measured'' in order to find out how
many characters have appeared since the beginning of the line or since the
most recent \<tab>.
\begintt
\def\setupverbatim{\tt \lineno=0
  \def\par{\leavevmode\egroup\box0\endgraf}
  \obeylines \uncatcodespecials \obeyspaces
  \catcode`\`=\active \catcode`\^^I=\active
  \everypar{\advance\lineno by1
    \llap{\sevenrm\the\lineno\ \ }\startbox}}
\newdimen\w \setbox0=\hbox{\tt\space} \w=8\wd0 % tab amount
\def\startbox{\setbox0=\hbox\bgroup}
{\catcode`\^^I=\active
  \gdef^^I{\leavevmode\egroup
    \dimen0=\wd0 % the width so far, or since the previous tab
    \divide\dimen0 by\w
    \multiply\dimen0 by\w % compute previous multiple of \w
    \advance\dimen0 by\w  % advance to next multiple of \w
    \wd0=\dimen0 \box0 \startbox}}
\endtt
^^|\divide|^^|\multiply|
(The new things in |\setupverbatim| are the `|\egroup\box0|' in the
redefinition of\/ |\par|; the `|\catcode`\^^I=\active|'; and the
`|\startbox|' in |\everypar|.) \ The |\settabs| and~|\+| macros of
Appendix~B provide another example of how tab operations can be simulated
by boxing and unboxing.

\smallbreak
\def\beginvrulealign{\setbox0=\vbox\bgroup}
\def\endvrulealign{\egroup % now \box0 holds the alignment}
  \setbox0=\vbox{\setbox2=\hbox{\vrule height\ht0 depth\dp0 width0pt}
    \unvbox0 \setbox0=\lastbox % now \box0 is the bottom row
    \nointerlineskip \copy0 % put it back
    \global\setbox1=\hbox{}
    \setbox4=\hbox{\unhbox0
      \loop \skip0=\lastskip \unskip % remove tabskip glue
       \advance\skip0 by-.4pt % rules are .4pt wide
       \divide\skip0 by 2
       \global\setbox1=\hbox{\hskip\skip0\vrule\hskip\skip0
         \unhbox2\unhbox1}%
       \setbox2=\lastbox % remove alignment entry
       \ifhbox2 \setbox2=\hbox{\kern\wd2}\repeat}}%
  \hbox{\rlap{\box0}\box1}}
\setbox8=\vbox{
\beginvrulealign
\tabskip=10pt
\halign{&\strut#\hfil\cr
These&    after\cr
vertical& the\cr
rules&    alignment\cr
were&     was\cr
inserted& completed!\cr}
\endvrulealign
}
Chapter 22 explains how to put vertical ^{rules in tables} by considering the
rules to be separate columns. There's also another way, provided that
the rules extend all the way from the top of the table to the bottom.
For example,
\begindisplay
|\beginvrulealign|\cr
|\tabskip=10pt|\cr
|\halign{&\strut#\hfil\cr|\cr
|These&    after\cr|\cr
|vertical& the\cr|\kern70pt yields\qquad\smash{$\vcenter{\box8}$}\cr
|rules&    alignment\cr|\cr
|were&     was\cr|\cr
|inserted& completed!\cr}|\cr
|\endvrulealign|\cr
\enddisplay
The magic macros in this case examine the bottom row of the
^{alignment}, which consists of alternating ^{tabskip glue} and boxes;
each item of tabskip glue in that bottom row will be bisected by
a vertical rule. Here's how:
\beginlines
|\def\beginvrulealign{\setbox0=\vbox\bgroup}|
|\def\endvrulealign{\egroup % now \box0 holds the entire alignment|
|  \setbox0=\vbox{\setbox2=\hbox{\vrule height\ht0 depth\dp0 width0pt}|
|    |^|\unvbox||0 \setbox0=\lastbox % now \box0 is the bottom row|
|    \nointerlineskip \copy0 % put it back|
|    \global\setbox1=\hbox{} % initialize box that will contain rules|
|    \setbox4=\hbox{\unhbox0 % now open up the bottom row|
|      \loop \skip0=|^|\lastskip|| |^|\unskip|| % remove tabskip glue|
|       \advance\skip0 by-.4pt % rules are .4pt wide|
|       \divide\skip0 by 2|
|       \global\setbox1=\hbox{\hskip\skip0\vrule\hskip\skip0|
|         \unhbox2\unhbox1}%|
|       \setbox2=|^|\lastbox|| % remove alignment entry|
|       \ifhbox2 \setbox2=\hbox{\kern\wd2}\repeat}}%|
|  \hbox{\rlap{\box0}\box1}} % superimpose the alignment on the rules|
\endlines
This method works with all alignments created by ^|\halign||{...}|. For
alignments created by, say, |\halign to100pt{...}|, the method works
only if the bottom row of the alignment contains all of the columns,
and only if `|\box1|' is replaced by `|\hbox to100pt{\unhbox1}|'
at the end of\/ |\endvrulealign|.

\begingroup
\eightpoint
\hyphenpenalty10000 \exhyphenpenalty10000 \pretolerance10000 % no hyphens
\newbox\dbox \setbox\dbox=\hbox to .4em{\hss.\hss} % dot box for leaders
\newskip\rrskipb \rrskipb=.5em plus3em % ragged right space before break
\newskip\rrskipa \rrskipa=-.17em plus-3em minus.11em % ditto, after
\newskip\rlskipa \rlskipa=0pt plus3em % ragged left space after break
\newskip\rlskipb \rlskipb=.33em plus-3em minus .11em % ditto, before
\newskip\lskip \lskip=3.3\wd\dbox plus1fil minus.3\wd\dbox % for leaders
\newskip\lskipa \lskipa=-2.67em plus-3em minus.11em % after leaders
\mathchardef\rlpen=1000 \mathchardef\leadpen=600 % constants used
\def\rrspace{\nobreak\hskip\rrskipb\penalty0\hskip\rrskipa}
\def\rlspace{\penalty\rlpen\hskip\rlskipb\vadjust{}\nobreak\hskip\rlskipa}
\uccode`~=` \uppercase{
  \def\:{\nobreak\hskip\rrskipb \penalty\leadpen \hskip\rrskipa
    \vadjust{}\nobreak\leaders\copy\dbox\hskip\lskip
    \kern3em \penalty\leadpen \hskip\lskipa
    \vadjust{}\nobreak\hskip\rlskipa \let~=\rlspace}
  \everypar{\hangindent=1.5em \hangafter=1 \let~=\rrspace}}
\parindent=0pt
\parfillskip=0pt
\obeyspaces
\def\text{ACM Symposium on Theory of Computing, Eighth Annual (Hershey, %
Pa., 1976)\:1879, 4813, 5414, 6918, 6936, 6937, 6946, 6951, %
6970, 7619, 9605, 10148, 11676, 11687, 11692, 11710, 13869}
\catcode`\ =10
\def\test#1 {\vbox{\hsize=#1em\hrule\strut\text\strut\par\hrule}}
\global\setbox5=\test 12.5
\global\setbox7=\test 17.5
%\global\setbox9=\test 22.5
\global\setbox9=\test 30
\endgroup
\subsection Paragraph maneuvers. Chapter 14 promised that Appendix D would
present an example where ^{ragged right} and ^{ragged left} setting
occur in the same paragraph. The following interesting example was
suggested by the ``^{Key Index}'' in {\sl ^{Mathematical Reviews}},
^^{index macros} where the entries consist of a possibly long title
followed by dot ^{leaders} followed by a possibly long list of
review numbers. If the title doesn't fit on one line, it should be set
ragged right, with hanging indentation on all lines after the first;
if the references don't all fit on one line, they should be set
ragged left. For example, given the input
\begintt
ACM Symposium on Theory of Computing, Eighth Annual (Hershey, %
Pa., 1976)\:1879, 4813, 5414, 6918, 6936, 6937, 6946, 6951, %
6970, 7619, 9605, 10148, 11676, 11687, 11692, 11710, 13869
\endtt
the following three types of output are desired, depending on the
column width:
\begindisplay
\box5&\box7\cr
\noalign{\vskip6pt}
\box9\span\cr
\enddisplay
Notice that the dot leaders are treated in three different ways, depending
on which works out best: They may occur at the left of the first line after
the title, or they may appear at the end of the last line of the title
(in which case they stop well before the right margin), or they may occur
in the middle of a line. Furthermore, the ragged-right lines are supposed
to end at least $0.5\em$ from the right margin. Our goal is to achieve
all this as a special case of \TeX's general paragraphing method. The
simple approach of Appendix~B won't work, because |\raggedright| is
achieved there by adjusting ^|\rightskip|; \TeX\ uses the same |\rightskip|
value in all lines of a paragraph.

The solution to this problem requires an understanding of the ^{line-breaking}
algorithm; it depends on how demerits are calculated, and on how items
are removed at the breakpoints, so the reader should review Chapter~14
until those concepts are firmly understood. Basically, we need to
specify a sequence of box/glue/penalty items for the spaces in the
title portion, another sequence for the spaces in the reference portion,
and another sequence for the dot leaders. In the title portion of each
index entry, interword spaces can be represented by the sequence
\begintt
\penalty10000 \hskip.5em plus3em \penalty0
\hskip-.17em plus-3em minus.11em
\endtt
Thus, there is a stretchability of $3\em$ if a line break occurs
at the |\penalty0|; otherwise the net interword space will be $.33\em$,
shrinkable to $.22\em$. This gives ragged right margins. The interword
spaces in the reference portion are designed to produce ragged left margins
and to minimize the number of lines devoted to references:
\begintt
\penalty1000 \hskip.33em plus-3em minus.11em
\vadjust{}\penalty10000 \hskip0pt plus3em
\endtt
The ^|\vadjust||{}| does nothing, but it doesn't disappear at a line break.
Thus, if a break occurs at the |\penalty1000|, the following line will
begin with stretchability $3\em$; but if no break occurs, the net space
will be $.33\em$ minus $.11\em$. Finally, the transition between title
and references can be specified by
\begintt
\penalty10000 \hskip.5em plus3em \penalty600
\hskip-.17em plus-3em minus.11em
\vadjust{}\penalty10000
\leaders\copy\dbox\hskip3.3\wd\dbox plus1fil minus.3\wd\dbox
\kern3em \penalty600 \hskip-2.67em plus-3em minus.11em
\vadjust{}\penalty10000 \hskip0pt plus3em
\endtt
(Quite a mouthful.) \
This long sequence of penalty and glue items begins rather like the interword
spaces in the first part, and it ends rather like the interword spaces in
the last part. It has two permissible breakpoints, namely at the
`|\penalty600|' items. The first breakpoint causes the leaders to
appear at the beginning of a line; the second causes them to appear
at the end, but 3~ems away. The leader width will always be at least
three times the width of\/ |\dbox|, so at least two copies of\/ |\dbox|
will always appear. Here is the actual \TeX\ code that can be used
to set up the desired behavior:
\beginlines
|\hyphenpenalty10000 \exhyphenpenalty10000 |^|\pretolerance||10000 % no hyphens|
^|\newbox||\dbox \setbox\dbox=\hbox to .4em{\hss.\hss} % dot box for leaders|
^|\newskip||\rrskipb \rrskipb=.5em plus3em % ragged right space before break|
|\newskip\rrskipa \rrskipa=-.17em plus-3em minus.11em % ditto, after|
|\newskip\rlskipa \rlskipa=0pt plus3em % ragged left space after break|
|\newskip\rlskipb \rlskipb=.33em plus-3em minus .11em % ditto, before|
|\newskip\lskip \lskip=3.3\wd\dbox plus1fil minus.3\wd\dbox % for leaders|
|\newskip\lskipa \lskipa=-2.67em plus-3em minus.11em % after leaders|
^|\mathchardef||\rlpen=1000 \mathchardef\leadpen=600 % constants used|
|\def\rrspace{|^|\nobreak||\hskip\rrskipb\penalty0\hskip\rrskipa}|
|\def\rlspace{|%
  |\penalty\rlpen\hskip\rlskipb\vadjust{}\nobreak\hskip\rlskipa}|\kern-2pt
|\uccode`~=` \uppercase{|
|  \def\:{\nobreak\hskip\rrskipb \penalty\leadpen \hskip\rrskipa|
|    \vadjust{}\nobreak\leaders\copy\dbox\hskip\lskip|
|    \kern3em \penalty\leadpen \hskip\lskipa|
|    \vadjust{}\nobreak\hskip\rlskipa \let~=\rlspace}|
|  \everypar{\hangindent=1.5em \hangafter=1 \let~=\rrspace}}|
|\uccode`~=0 |^|\parindent||=0pt |^|\parfillskip||=0pt |^|\obeyspaces|
\endlines
Putting the interword glue into |\skip| registers saves a great deal
of time and memory space when \TeX\ works with such paragraphs;
`|\hskip|\<explicit glue>' occupies six cells of \TeX's ^{box memory},
but `|\hskip|\<skip register>' occupies only two. ^^{efficiency}
Notice the tricky use of\/ ^|\uppercase| here to convert |~|$_{13}$ into
\]$_{13}$; ^^{active spaces} ``random'' ^{active characters} can be
obtained in a similar way.

\medbreak
Let's turn now to a much simpler problem: {\sl^{hanging punctuation}}.
\medskip
\setbox0=\vbox{\vfil
\hyphenpenalty=10000
\eightpoint
\hsize=14pc
\newdimen\commahang \setbox0=\hbox{,} \commahang=\wd0
\newdimen\periodhang \setbox0=\hbox{.} \periodhang=\wd0
\newdimen\quotehang \setbox0=\hbox{`} \quotehang=\wd0
\newdimen\qquotehang \setbox0=\hbox{``} \qquotehang=\wd0
\newskip\zzz \def\allowhyphens{\nobreak\hskip\zzz}
\def\lqq{``} \def\rqq{''} \def\pnt{.}
\def\comma{,\kern-\commahang\kern\commahang}
\def\period{.\kern-\periodhang\kern\periodhang}
\def\rquote{'\kern-\quotehang\kern\quotehang}
\def\lquote{\ifhmode\kern\quotehang\vadjust{}\else\leavevmode\fi
  \kern-\quotehang`\allowhyphens}
\catcode`,=\active \let,=\comma
\catcode`.=\active \let.=\period
\catcode`'=\active \def'{\futurelet\next\rqtest}
\catcode``=\active \def`{\futurelet\next\lqtest}
\def\rqtest{\ifx\next'\let\next=\rquotes\else\let\next=\rquote\fi\next}
\def\lqtest{\ifx\next`\let\next=\lquotes\else\let\next=\lquote\fi\next}
\def\rquotes'{\rqq\kern-\qquotehang\kern\qquotehang}
\def\lquotes`{\ifhmode\kern\qquotehang\vadjust{}\else\leavevmode\fi
  \kern-\qquotehang\lqq\allowhyphens}
\parindent=0pt
``What is hanging punctuation?'' asked Alice, with a puzzled frown.
`Well, y'know, actually,' answered Bill, `I'd rather demonstrate it than
explain it.' ``Oh, now I see.  Commas, periods, and quotes are allowed to
stick out into the margins, if they occur next to a line break.'' `Yeah, I
guess.' ``Really! But why do all your remarks have single quotes, while
mine are double?'' `I haven't the foggiest; it's weird. Ask the author of
this crazy book.'
}\setbox2=\vsplit0 to.5\ht0 \line{\box2\hfil\box0}
\medskip\noindent
Each comma in Alice and Bill's demonstration paragraph
was represented inside of \TeX\ by
the sequence of three items `|,\kern-\commahang|^|\kern||\commahang|',
and there were similar replacements for periods and for closing quotes;
opening quotes were represented by the longer sequence
\begintt
\kern\qquotehang\vadjust{}\kern-\qquotehang``\allowhyphens
\endtt
where ^|\allowhyphens| allows the following word to be ^{hyphenate}d.
This construction works because kerns disappear into ^{line breaks} in the
proper way; the relevant rules from Chapter~14 are: (1)~A line break can
occur at a kern that is immediately followed by glue. (2)~Consecutive
glue, kern, and penalty items disappear at a break.

To set \TeX\ up for hanging punctuation, you can say
\beginlines
|\newdimen\commahang      \setbox0=\hbox{,}      \commahang=\wd0|
|\newdimen\periodhang     \setbox0=\hbox{.}      \periodhang=\wd0|
|\newdimen\quotehang      \setbox0=\hbox{`}      \quotehang=\wd0|
|\newdimen\qquotehang     \setbox0=\hbox{``}     \qquotehang=\wd0|
|\newskip\zzz \def\allowhyphens{\nobreak\hskip\zzz}|
|\def\lqq{``} \def\rqq{''} \def\pnt{.}|
|\def\comma{,\kern-\commahang\kern\commahang}|
|\def\period{.\kern-\periodhang\kern\periodhang}|
|\def\rquote{'\kern-\quotehang\kern\quotehang}|
|\def\lquote{\ifhmode\kern\quotehang\vadjust{}\else\leavevmode\fi|
|  \kern-\quotehang`\allowhyphens}|
|\catcode`,=\active \let,=\comma \catcode`.=\active \let.=\period|
|\catcode`'=\active \def'{\futurelet\next\rqtest}|
|\catcode``=\active \def`{\futurelet\next\lqtest}|
|\def\rqtest{\ifx\next'\let\next=\rquotes\else\let\next=\rquote\fi\next}|
|\def\lqtest{\ifx\next`\let\next=\lquotes\else\let\next=\lquote\fi\next}|
|\def\rquotes'{\rqq\kern-\qquotehang\kern\qquotehang}|
|\def\lquotes`{\ifhmode\kern\qquotehang\vadjust{}\else\leavevmode\fi|
|  \kern-\qquotehang\lqq\allowhyphens}|
\endlines
Notice that the macros need to do their own checking for ligatures, and
they also take appropriate actions when a paragraph begins with an opening
quote.  Since |\kern| does not affect the ^{space factor}, hanging
punctuation doesn't affect \TeX's spacing conventions within a line.
Partially hanging punctuation can be obtained by decreasing the amounts of\/
|\commahang|, etc.  The macros ^|\pnt|, ^|\lq|, and ^|\rq| should be used
in constants; for example, a dimension of $6.5\,$in must be written
`|6\pnt5in|' when hanging punctuation is in effect, and
`|\catcode\lq,=12|' makes commas inactive again. A special font with
zero-width ^|\hyphenchar| should be used for ``hanging hyphenation.''

\medbreak
And now for our next trick, let's consider an application to short footnotes.
^^{footnotes, short} The footnotes at the bottom of this page%
\footnote{$^1$}{First footnote.}$^,$%
\footnote{$^2$}{Second footnote. (Every once in a~while a long
  footnote might occur, just to make things difficult.)}$^,$%
\footnote{$^3$}{Third footnote.}$^,$%
\footnote{$^4$}{Fourth footnote.}$^,$%
\footnote{$^5$}{Fifth footnote. (This is incredibly boring,
  but it's just an example.)}$^,$%
\footnote{$^6$}{Another.}$^,$%
\footnote{$^7$}{And another.}$^,$%
\footnote{$^8$}{Ho hum.}$^,$%
\footnote{$^9$}{Umpteenth footnote.}$^,$%
\footnote{$^{10}$}{Oodles of them.}
look funny, because most of them are quite short. When a document has lots
of footnotes, and when most of them take up only a small part of a line,
the output routine ought to reformat them in some more appropriate way.

\newcount\footno
For example, one approach would be to typeset the footnotes in narrow
columns and to put, say, three columns of footnotes at the bottom of
each page. The ten example footnotes might then look like this:
\medskip
\begingroup
\def\vfootnote#1{\insert\footins{
    \eightpoint \hsize=9pc \parindent=1pc
    \leftskip=0pt \raggedright \pretolerance=10000
    \interlinepenalty=\interfootnotelinepenalty
    \floatingpenalty=20000
    \splittopskip=\ht\strutbox \splitmaxdepth=\dp\strutbox
    \global\advance\footno by 1
    \item{$^{\the\footno}$}\strut#1\strut\par\allowbreak}}
\output={\showboxbreadth=9999 \showboxdepth=1
  \nonstopmode \showbox\footins \errorstopmode
  \rigidbalance\footins 3 7pt
  \global\setbox1=\lastbox
  \unvbox255}
\footno=0
\vfootnote{First footnote.}
\vfootnote{Second footnote. (Every once in a~while a long
  footnote might occur, just to make things difficult.)}
\vfootnote{Third footnote.}
\vfootnote{Fourth footnote.}
\vfootnote{Fifth footnote.
  (This is incredibly boring, but it's just an example.)}
\vfootnote{Another.}
\vfootnote{And another.}
\vfootnote{Ho hum.}
\vfootnote{Umpteenth footnote.}
\vfootnote{Oodles of them.}
\newcount\k \newdimen\h
\def\rigidbalance#1#2 #3 {\setbox0=\box#1 \k=#2 \h=#3
  \line{\splittopskip=\h \vbadness=10000 \hfilneg
  \valign{##\vfil\cr\dosplits}}}
\def\dosplits{\ifnum\k>0 \noalign{\hfil}\splitoff
  \global\advance\k-1\cr\dosplits\fi}
\def\splitoff{\dimen0=\ht0 \divide\dimen0 by\k \advance\dimen0 by\h
  \vsplit0 to \dimen0 }
\eject
\box1\endgroup
\medskip\noindent
In this case, the footnotes could be generated by
\begindisplay
|\insert\footins{\eightpoint \hsize=9pc \parindent=1pc|\cr
|  \leftskip=0pt |^|\raggedright|| \pretolerance=10000|\cr
|  \hyphenpenalty=10000 \exhyphenpenalty=10000|\cr
|  \interlinepenalty=\interfootnotelinepenalty|\cr
|  \floatingpenalty=20000|\cr
|  \splittopskip=\ht\strutbox \splitmaxdepth=\dp\strutbox|\cr
|  \item{$^{\the\footno}$}\strut|\<text of footnote>^|\strut|\cr
|  \par|^|\allowbreak||}|\cr
\enddisplay
and |\count\footins| would be set to 333 so that each footnote line
would be considered to occupy about one third of a line on the page.
The output routine would then see a |\box\footins| that looks like
this:
\begindisplay
|\vbox(142.0+2.0)x108.0|\cr
|.\hbox(7.0+2.0)x108.0, glue set 42.23425fil []|\cr
|.\penalty 0|\cr
|.\hbox(7.0+2.0)x108.0, glue set 0.29266 []|\cr
|.\penalty 250|\cr
|.\glue(\baselineskip) 1.44444|\cr
|.\hbox(5.55556+1.55556)x96.0, glue set 0.8693, shifted 12.0 []|\cr
|.\penalty 100|\cr
|.\glue(\baselineskip) 1.88889|\cr
|.\hbox(5.55556+1.55556)x96.0, glue set 0.92438, shifted 12.0 []|\cr
\qquad\vdots\cr
|.\hbox(7.0+2.0)x108.0, glue set 18.56308fil []|\cr
|.\penalty 0|\cr
|.\hbox(7.0+2.0)x108.0, glue set 36.92476fil []|\cr
|.\penalty 0|\cr
\enddisplay
The individual footnotes each end with `|\penalty 0|'; footnotes that
take up more than one line have larger penalties between the lines,
and interline glue appears there too.

How should the output routine break such a box up into three roughly
equal pieces? Notice that the contents of the box are completely rigid;
i.e., there is no glue that can stretch or shrink. Furthermore, we can
assume that the contents of the box are regular, i.e., that the
interbaseline distances are all the same. In such circumstances
a fairly simple ^{balancing} routine can be used to trisect the box.

Let's consider a more general problem: Suppose that a rigid vbox is given,
$n$~lines tall, where adjacent baselines are $b$~units apart. Suppose
also that the top baseline is $h$~units from the top of the
vbox, where $0<h<b$. \ (In our footnote example, $b=9\pt$ and $h=7\pt$;
in the standard settings of plain \TeX, $b=12\pt$ and $h=10\pt$. We might
as well work the problem for general $b$ and $h$.) \ It follows that
the height of the vbox is $H=h+b(n-1)=bn+h-b$.

If $n$ lines are to be distributed evenly into $k$ columns, the first
column should contain $\lceil n/k\rceil$ lines. \ (This denotes the
smallest integer greater than or equal to $n/k$.) \ For example,
our application to footnotes has $n=16$ and $k=3$, hence the first
column should contain 6~lines. After forming the first column,
we have reduced the problem to $n=10$ and $k=2$, so two 5-line
columns will complete the operation. \ (Notice that it is better
to divide 16 into $6+5+5$ instead of $6+6+4$.) \ Once we have
found the first column, it's always possible to reduce the $k$-column
problem to a $(k-1)$-column problem, so we need only concentrate
on finding the first column.

Let $m=\lceil n/k\rceil$. The height of the given box is $bn+h-b$,
and the height of the first column should be $bm+h-b$; hence
we want to do a ^|\vsplit| to that height. However, it isn't
necessary to calculate $bm+h-b$ exactly, since a bit of
arithmetic proves that
\begindisplay
$\displaystyle bm+h-b\;<\;{bn+h-b\over k}+h\;<\;b(m+1)+h-b.$
\enddisplay
Therefore it suffices to |\vsplit| to height $H'=H/k+h$; under the
assumptions of rigidity, and assuming that a valid break is possible
after each line, |\vsplit| to $H'$ will split after the maximum
number of lines that yield a box of height $\le H'$. \ (We have
observed that $m$~lines produce a box of height~$<H'$ while
$m+1$~lines produce a box of height~$>H'$.) \ The following \TeX\ code
does this:
\beginlines
|\newcount\k \newdimen\h % registers used by the \rigidbalance routine|
|\def\rigidbalance#1#2 #3 {\setbox0=\box#1 \k=#2 \h=#3|
|  \line{\splittopskip=\h \vbadness=10000 \hfilneg|
|    \valign{##\vfil\cr\dosplits}}}|
|\def\dosplits{\ifnum\k>0 \noalign{\hfil}\splitoff|
|  \global\advance\k-1\cr\dosplits\fi}|
|\def\splitoff{\dimen0=\ht0|
|  |^|\divide||\dimen0 by\k \advance\dimen0 by\h|
|  \vsplit0 to \dimen0 }|
\endlines
This code is interesting on a number of counts. First, notice that
the calculation does not depend on~$b$, only on~$h$ and the height of
the given box; hence |\rigidbalance| has three parameters: a box register
number, the number of columns~$k$, and the top baseline height~$h$.
The routine splits the given vbox into $k$~nearly equal pieces and
justifies the result in a |\line|. The value of\/ ^|\splittopskip| is
set to~$h$ so that subsequent vboxes will satisfy the ground rules of the
original vbox, as the problem is reduced from~$k$ to~$k-1$. Each column will
be preceded by ^|\hfil|, hence ^|\hfilneg| is used to cancel the |\hfil|
before the first column. A ^|\valign| is used to align all of the columns
at the top. Notice that the preamble to this |\valign| is
quite simple; and the body of the |\valign| is generated by a ^{recursive
macro} |\dosplits| that produces the $k$~columns. The value of\/
^|\vbadness| is set to~10000 because each |\vsplit| operation
will produce an ^{underfull} vbox whose badness is 10000.

In our application to footnotes, the |\output| routine can reformat
the contents of\/ |\box\footins| by saying, for example,
\begintt
\rigidbalance\footins 3 7pt
\setbox2=\lastbox \setbox\footins=\vbox{\box2}
\endtt
since ^|\lastbox| will be the result of\/ |\rigidbalance|, which is an hbox.

This solution to the problem of short footnotes might result in
^{club lines} or ^{widow lines}, since the balancing routine we have
described simply trisects the total number of lines. For example,
if the tenth footnote of our example had not been present,
the fifteen remaining lines would have been split $5+5+5$;
the second column would have been headed by the lonely
word `{\eightrm difficult.)}', and the third column would have
started with `{\eightrm just an example.)}'. The rigid balancing
procedure could be replaced by one that allows ^{ragged-bottom}
columns, but there's also another approach: The entire set of footnotes
could be combined into a single paragraph, with generous spacing
between the individual items. For example, the ten footnotes we
have been considering might appear as follows:
\medskip
\begingroup
{\catcode`p=12 \catcode`t=12 \gdef\\#1pt{#1}}
\let\getfactor=\\
\eightpoint
\newskip\footglue \footglue=1.5em plus.3em minus.3em
\newdimen\footnotebaselineskip \footnotebaselineskip=10pt
\dimen0=\footnotebaselineskip \multiply\dimen0 by 1024
\divide \dimen0 by \hsize \multiply\dimen0 by 64
\xdef\fudgefactor{\expandafter\getfactor\the\dimen0 }
\def\vfootnote#1{\insert\footins{\floatingpenalty=20000
  \global\advance\footno by 1
  \eightpoint \setbox0=\hbox{%
    $^{\the\footno}$#1\penalty-10\hskip\footglue}
  \dp0=0pt \ht0=\fudgefactor\wd0 \box0}}
\output={\showboxbreadth=9999 \showboxdepth=1
  \nonstopmode \showbox\footins \errorstopmode
  \global\setbox1=\vbox{\makefootnoteparagraph}
  \unvbox255}
\footno=0
\vfootnote{First footnote.}
\vfootnote{Second footnote. (Every once in a~while a long
  footnote might occur, just to make things difficult.)}
\vfootnote{Third footnote.}
\vfootnote{Fourth footnote.}
\vfootnote{Fifth footnote.
  (This is incredibly boring, but it's just an example.)}
\vfootnote{Another.}
\vfootnote{And another.}
\vfootnote{Ho hum.}
\vfootnote{Umpteenth footnote.}
\vfootnote{Oodles of them.}
%\def\makefootnoteparagraph{\unvbox\footins \makehboxofhboxes
% \showboxbreadth=9999 \showboxdepth=1
% \nonstopmode \showbox0 \errorstopmode
%  \setbox0=\hbox{\unhbox0 \removehboxes}
% \showboxbreadth=30
% \nonstopmode \showbox0 \errorstopmode
%  \baselineskip=10pt\noindent\unhbox0 }
%\def\makehboxofhboxes{\setbox0=\hbox{}
%  \loop\setbox2=\lastbox
%  \ifhbox2 \setbox0=\hbox{\box2\unhbox0}\repeat}
%\def\removehboxes{\setbox0=\lastbox
%  \ifhbox0{\removehboxes}\unhbox0 \fi}
\def\makefootnoteparagraph{\unvbox\footins
 \baselineskip=\footnotebaselineskip \removehboxes}
\def\removehboxes{\unskip\setbox0=\lastbox
 \ifhbox0{\removehboxes}\unhbox0 \else\noindent \fi}
\eject
\box1\endgroup
\medskip

It would be possible to take the contents of\/ |\box\footins| shown
previously and to reformat everything into a paragraph, but such an operation
would be needlessly complicated. If footnotes are to be paragraphed by the
output routine, it's better simply to prepare them in unjustified
hboxes. Each of these hboxes will be unboxed later, so we are free to play
with their heights, widths, and depths. It's convenient to set the
depth to zero and the height to an estimate of how much a particular
footnote will contribute to the final paragraph. For example, if
a footnote takes up exactly half of the |\hsize|, and if the final
footnote is going to be set with |\baselineskip=10pt|, then the
height of the footnote hbox should be set to $5\pt$. By letting
|\count\footins=1000|, we'll have a pretty good estimate of the size
of the final footnote paragraph. In other words, the following
insertion scheme is suggested:
\begindisplay
|\insert\footins{\floatingpenalty=20000|\cr
|  \eightpoint \setbox0=\hbox{%|\cr
|    $^{\the\footno}$|\<text of footnote>|\penalty-10\hskip\footglue}|\cr
|  \dp0=0pt \ht0=\fudgefactor\wd0 \box0}|\cr
\enddisplay
The penalty of $-10$ tends to favor line breaks between footnotes;
|\footglue| is the amount of glue between footnotes in the final
footnote paragraph; and |\fudgefactor| is the ratio of\/ |\baselineskip|
to |\hsize| in that paragraph. The author defined the necessary quantities
as follows in his experiments:
\begintt
\eightpoint \newskip\footglue \footglue=1.5em plus.3em minus.3em
\newdimen\footnotebaselineskip \footnotebaselineskip=10pt
\dimen0=\footnotebaselineskip \multiply\dimen0 by 1024
\divide \dimen0 by \hsize \multiply\dimen0 by 64
\xdef\fudgefactor{\expandafter\getfactor\the\dimen0 }
\endtt
^^|\multiply| ^^|\divide| ^^|sp|
(The computation of\/ |\fudgefactor| uses the fact that $1\pt=1024\times64\,$sp,
and it assumes that the |\footnotebaselineskip| is less than $16\pt$.)

Inside the output routine, |\box\footins| will now be a vbox of hboxes, and the
height of this vbox will be an estimate of the height of the final paragraph.
For example, our ten footnotes produce
\begintt
\vbox(34.48158+0.0)x386.4221
.\hbox(2.00175+0.0)x70.68285 []
.\hbox(10.94359+0.0)x386.4221 []
.\hbox(2.09749+0.0)x74.06345 []
.\hbox(2.2077+0.0)x77.95517 []
.\hbox(7.6296+0.0)x269.40376 []
.\hbox(1.40851+0.0)x49.73532 []
.\hbox(1.87659+0.0)x66.26334 []
.\hbox(1.38826+0.0)x49.02003 []
.\hbox(2.67213+0.0)x94.35402 []
.\hbox(2.25597+0.0)x79.65926 []
\endtt
and the height of $34.48158\pt$ corresponds to an estimate of about
three and a~half lines. \ (\TeX's page builder has also added |\skip\footins|
when estimating the total contribution due to footnotes.)

Finally, the reformatting of\/ |\box\footins| can be achieved easily with
an elegant technique suggested by David ^{Kastrup}, using the following
\TeX\ code within the |\output| routine:
\begindisplay
|\def\makefootnoteparagraph{\unvbox\footins|\cr
|  \baselineskip=\footnotebaselineskip \removehboxes}|\cr
|\def\removehboxes{\unskip\setbox0=\lastbox|\cr
|  \ifhbox0{\removehboxes}\unhbox0 \else\noindent \fi}|\cr
\enddisplay
^^|\unhbox|^^|\unvbox|
The key idea here is |\removehboxes|, a macro that has the magical ability to
take a vertical box such as `|\vbox{\box1\box2\box3\removehboxes}|' and
transform it into
`|\vbox{\noindent\unhbox1\unhbox2\unhbox3}|'\kern-1pt,
if\/ |\box1|, \kern-2pt|\box2|, and
|\box3| are hboxes. Notice how |\removehboxes| introduces braces so that
\TeX's ^{save stack} will hold all of the hboxes before they are unboxed. Each
level of recursion in this routine uses one cell of input stack space and
three cells of save stack space; thus, it is generally safe to do more than
100 footnotes without exceeding \TeX's capacity.

In our application there is no ^{interline glue} within |\box\footins|,
so the ^|\unskip| command could be deleted from |\removehboxes|.

Incidentally, the |\unskip| and ^|\lastbox| operations have running
times of the approximate form
$a+mb$, where $m$~is the number of items ^^{efficiency} on the
list preceding the glue or box that is removed. Hence |\removehboxes| has a
running time of order $n^2$ when it removes $n$~boxes.
But the constant~$b$ is so small that
for practical purposes it's possible to think of\/ |\unskip| and
|\lastbox| as almost instantaneous.

\subsection Communication with output routines. It would be possible to
write an entire book about \TeX\ output routines; but the present
appendix is already too long, so it will suffice to mention only one
or two sneaky tricks that a person might not readily think~of. \ (Appendix~E
gives some less sneaky examples.)

Sometimes an output routine needs to know why it was invoked, so there's
a problem of communicating information from the rest of the program.
\TeX\ provides general |\mark| operations, but marks don't always yield
the right sorts of clues. Then there's ^|\outputpenalty|, which can be
tested to see what penalty occurred at a breakpoint; any penalty of
$-10000$, $-10001$, $-10002$, or less,
forces the output routine to act, hence different penalty values
can be used to pass different messages. \ (When the output routine puts
material back on the list of contributions, it need not restore the
penalty at the breakpoint.) \ If output has been forced by a
highly negative value of\/ |\outputpenalty|, the output routine can
use |\vbox{\unvcopy255}| to discover how full the page-so-far actually is.
Underfull and overfull boxes are not reported when |\box255| is
packaged for use by the output routine, so there's no harm in ejecting
a page prematurely if you want to pass a signal. \ (Set
^|\holdinginserts| positive to pass a signal when the contents of\/
|\box255| will be sent back through the page builder again, if any
insertions are present.)

Perhaps the dirtiest trick of all is to communicate with the output
routine via the {\sl depth\/} of\/ |\box255|. For example, suppose that
you want to know whether or not the current page ends with the last
line of a paragraph. If each paragraph ends with `|\specialstrut|', where
|\specialstrut| is like ^|\strut| but $1\,$sp deeper, then |\dp255|
will have a recognizable value if a page ends simultaneously
with a paragraph. \ (Of course, ^|\maxdepth| must be suitably large; plain
\TeX\ takes |\maxdepth=4pt|, while struts are normally $3.5\pt$ deep, so
there's no problem.) \ A distance of~$1000\,$sp is invisible to the
naked eye, so a variety of messages can be passed in this way.

If the value of\/ ^|\vsize| is very small, \TeX\ will construct paragraphs
as usual but it will send them to the output routine one line at a time.
In this way the output routine could attach marginal notes, etc., based
on what occurs in the line. Paragraphs that have been rebuilt in this
way can also be sent back from the output routine to the page builder;
normal page breaks will then be found, if\/ |\vsize| has been restored.

An output routine can also write notes on a file, based on what occurs in
a manuscript. A two-pass system can be devised where \TeX\ simply gathers
information during the first pass; the actual typesetting can be done during
the second pass, using |\read| to recover information that was written
during the first.

\subsection Syntax checking. Suppose you want to run a manuscript through
\TeX\ just to check for errors, without getting any output. Is there a
way to make \TeX\ run significantly faster while doing this? Yes; here's how:
(1)~Say `|\font|^|\dummy||=dummy|'\kern1pt; your system should include a
file |dummy.tfm| that defines a font with no characters (but with
enough |\fontdimen| parameters to qualify as a math symbol font).
(2)~Set all the font identifiers you are using equal to |\dummy|. For
example, |\let\tenrm=\dummy|, |\let\tenbf=\dummy|, \dots, |\textfont0=\dummy|,
etc. (3)~Say `|\dummy|' to select the dummy font (since plain \TeX\ may
have selected the real |\tenrm|).  (4)~Set ^|\tracinglostchars||=0|, so that
\TeX\ won't complain when characters aren't present in the dummy font. (5)~Set
\begintt
\output={\setbox0=\box255\deadcycles=0}
\endtt
^^|\deadcycles|
so that nothing will be shipped out, yet \TeX\ will not think that your
output routine is flaky. (6)~Say ^|\newtoks||\output|, so that no other
output routine will be defined. (7)~Say ^|\frenchspacing| so that \TeX\
will not have to do space factor calculations. (8)~Say
^|\hbadness||=10000| so that underfull boxes will not be reported.
(9)~And if you want to disable |\write| commands, use the following
trick due to Frank ^{Yellin}:
\begintt
\let\immediate=\relax \def\write#1#{{\afterassignment}\toks0=}
\endtt
These changes usually make \TeX\ run more than four times as fast.
^^|\afterassignment|

\endchapter

^{Wolfe}, who had moved around the desk and into his chair,
^^{Hombert, Humbert} put up a palm at him: ``Please, Mr.~Hombert.
I think it is always advisable to take a short-cut when it is feasible.\rlap{''}
\author REX ^{STOUT}, {\sl The Rubber Band\/} (1936)

\bigskip

``My dear ^{Watson}, try a little analysis yourself,\rlap{''}
said he, ^^{Holmes} with a touch of impatience.
``You know my methods. Apply them,
and it will be instructive to compare results.\rlap{''}
\author CONAN ^{DOYLE}, {\sl The Sign of the Four\/} (1890)

\eject
\beginchapter Appendix E. Example Formats

Although the plain \TeX\ format of Appendix B is oriented to technical
reports, it can readily be adapted to quite different applications. Examples of
three such adaptations are provided in this appendix: (1)~a~format for
business letters; (2)~a~format for concert programs; (3)~the~format used
to typeset this book.

Let's consider ^{business letters} first. Suppose that you want \TeX\ to
format your ^{correspondence}, and that you have $n$~letters to send. ^^{mail}
If your computer system contains a file ^|letterformat.tex| like the one
described later in this appendix, it's easy to do the job by applying \TeX\ to
a file that looks like this:
\begindisplay
\<optional magnification>\cr
^|\input|| letterformat|\cr
\<business letter$_1$>\cr
\noalign{\vskip-1pt}
\qquad\vdots\cr
\<business letter$_n$>\cr
^|\end|\cr
\enddisplay
Here each of the $n$ business letters has the form
\begindisplay
\<letterhead>\cr
^|\address|\cr
\<one or more lines of address>\cr
^|\body|\cr
\<one or more paragraphs of text>\cr
^|\closing|\cr
\<one or more lines for salutation and signature>\cr
\<optional annotations>\cr
\<optional postscripts>\cr
^|\endletter|\cr
^|\makelabel|| % omit this if you don't want an address label|\cr
\enddisplay
The \<letterhead> at the beginning of this construction is usually a
control sequence like |\rjdletterhead| for letters by R.~J.~D.; each
letter writer can have a personalized letterhead that is stored with the
|letterformat| macros. The \<optional annotations> at the end are any number
of one-line notes preceded by `^|\annotations|'; the \<optional postscripts>
are any number of paragraphs preceded by `^|\ps|'. When \TeX\ is processing the
|\address| and the |\closing| and the optional |\annotations|, it produces
output line-for-line just as the lines appear in the input file; but when
\TeX\ is processing the |\body| of the letter and the optional |\ps|, it
chooses line breaks and justifies lines as it normally does when
typesetting paragraphs in books.

A complete example, together with the resulting output, appears on the
next two pages. This example starts with `^|\magnification||=|^|\magstep||1|'
because the letter is rather short. Magnification is usually omitted if the
letters are long-winded; `|\magnification=|^|\magstephalf|' is
appropriate when they are medium-size. The same magnification applies to
all~$n$ letters, so you must run \TeX\ more than once if you want more than
one magnification.

\eject
\begingroup \obeylines
|\magnification=\magstep1|
|\input letterformat|
\bigskip
|\rjdletterhead % (see the output on the next page)|
\bigskip
|\address|
|Prof.~Brian~K. Reid|^^{Reid}
|Department of Electrical Engineering|
|Stanford University|
|Stanford, CA 94305|
\bigskip
|\body|
|Dear Prof.~Reid:|
\bigskip
|I understand that you are having difficulties with|
|Alka-Seltzer tablets.  Since there are 25~pills|^^{Alka-Seltzer}^^{ties}
|per bottle, while the manufacturer's directions|
|recommend ``plop,~plop, fizz,~fizz,'' my colleagues|
|tell me that you have accumulated a substantial|
|number of bottles in which there is one tablet|
|left. % (See the 1978 SCRIBE User Manual, page 90.)|^^{Scribe}
\bigskip
|At present I am engaged in research on the potential|
|applications of isolated analgesics. If you would|
|be so kind as to donate your Alka-Seltzer collection|
|to our project, I would be more than happy to send|
|you preprints of any progress reports that we may|
|publish concerning this critical problem.|
\bigskip
|\closing|
|Sincerely,|
|R. J. Drofnats|^^{Drofnats}
|Professor|
\bigskip
|\annotations|
|RJD/dek|
|cc: {\sl The \TeX book}|
\bigskip
|\ps|
|P. S. \ If you like, I will check into the|
|possibility that your donation and the meals that|
|you have been eating might be tax-deductible, in|^^{IRS}
|connection with our research.|
|\endletter|
|\makelabel|
\eject\endgroup
\def\proofcopy(#1){\ifproofmode\raise.5in\hbox{\sevenrm(#1)}\else\hfil\fi}
\vglue-\topskip \nointerlineskip
\dimen0=\vsize \advance\dimen0 by-1.2pt \advance\dimen0 by-2.1in
\vbox{\hrule
  \hbox{\vrule height \dimen0
    \hbox to 4.25in{\hss\proofcopy(Output page goes here, reduced 50\%.)\hss}
    \vrule}}
\nointerlineskip
\hrule
\line{\vrule height 2.1in
  \hss\proofcopy(Label and stamp go here, reduced 50\%.)\hss\vrule}
\hrule
\vskip 0pt plus .001pt minus .001pt % in case of rounding errors
\eject
If the letter is more than one page long, the addressee, date, and
page number will appear at the top of subsequent pages. For example,
the previous letter comes out as follows, if additional paragraphs are
added to the text:
\medskip
\hrule
\line{\vrule height 3.1127in
  \hfil\proofcopy(First page, reduced to 28.3\%.)\hfil\vrule\hfil
  \proofcopy(Second page, reduced to 28.3\%.)\hfil\vrule}
\hrule
\bigskip
\ninepoint
The macro package |letterformat.tex| that produces this format begins
with a simple macro that expands to the current ^{date}.
\beginlines
|\def|^|\today||{|^|\ifcase|^|\month|^|\or|
|  January\or February\or March\or April\or May\or June\or|
|  July\or August\or September\or October\or November\or December\fi|
|  |^|\space|^|\number|^|\day||, \number|^|\year||}|
\endlines

Then comes the specification of page layout, which is ``ragged'' at the
bottom. A rather large ^|\interlinepenalty| is used so that page
breaks will tend to occur between paragraphs.
\beginlines
^|\raggedbottom|
|\interlinepenalty=1000|
^|\hsize||=6.25truein|
^|\voffset||=24pt|
|\advance|^|\vsize|| by-\voffset|
^|\parindent||=0pt|
^|\parskip||=0pt|
^|\nopagenumbers|
^|\headline||={\ifnum|^|\pageno||>1|
|  \tenrm To \addressee\hfil\today\hfil Page |^|\folio|
|  \else\hfil\fi}|
\endlines

The contents of a letter are typeset either in ``line mode'' (obeying lines)
or in ``paragraph mode'' (producing paragraphs in ^{block style}). Control
sequences |\beginlinemode| and |\beginparmode| are defined to initiate these
modes; and another control sequence, |\endmode|, is defined and redefined so
that the current mode will terminate properly:
\beginlines
|\def\beginlinemode{\endmode|
|  |^|\begingroup|^|\obeylines||\def\endmode{\par|^|\endgroup||}}|
|\def\beginparmode{\endmode|
|  \begingroup\parskip=\medskipamount \def\endmode{\par\endgroup}}|
|\let\endmode=\par|
|\def\endletter{\endmode\vfill|^|\supereject||}|
\endlines

One of the chief characteristics of this particular business letter format
is a parameter called ^|\longindentation|, which is used to indent the
closing material, the date, and certain aspects of the letterhead. The
^|\address| macro creates a box that will be used both in the letter and in
the label on the envelope. If individual lines of the address exceed
|\longindentation|, they are broken, and hanging indentation is used for any
material that must be carried over.
\beginlines
|\newdimen\longindentation \longindentation=4|^|true||in|
|\newbox\theaddress|
|\def\address{\beginlinemode\getaddress}|
|{\obeylines|^|\gdef||\getaddress #1|
|  #2|
|  {#1\gdef\addressee{#2}%|
|    \global\setbox\theaddress=\vbox|^|\bgroup|^|\raggedright||%|
|    \hsize=\longindentation |^|\everypar||{|^|\hangindent||2em}#2|
|    \def\endmode{\egroup\endgroup |^|\copy||\theaddress |^|\bigskip||}}}|
\endlines
(Parameter |#2| to |\getaddress| ^^{parameters, delimited}
will be the contents of the line following |\address|, i.e., the
name of the addressee.)

The closing macros are careful not to allow a page break anywhere between the
end of the ^|\body| and the beginning of a ^|\ps|.
\beginlines
|\def\body{\beginparmode}|
|\def\closing{\beginlinemode\getclosing}|
|{\obeylines\gdef\getclosing #1|
|  #2|
|  {#1|^|\nobreak||\bigskip |^|\leftskip||=\longindentation #2|
|    \nobreak\bigskip\bigskip\bigskip % space for signature|
|    \def|
|    {|^|\endgraf||\nobreak}}}|
|\def\annotations{\beginlinemode\def\par{\endgraf\nobreak}\obeylines\par}|
|\def\ps{\beginparmode\nobreak|
|  \interlinepenalty5000\def\par{\endgraf\penalty5000 }}|
\endlines

The remaining portion of |letterformat.tex| deals with ^{letterheads} and
labels, which of course will be different for different organizations.
The following macros were used to generate the examples in this
appendix; they can be modified in more-or-less obvious ways to produce
suitable letterheads of other kinds. Special fonts are generally
needed, and they should be loaded at `^|true|' sizes so that they are not
affected by magnification.  One tiny refinement worth noting here is the
^|\up| macro, which raises ^{brackets} so that they look better in a
^{telephone number}.
\beginlines
|\def\up#1{|^|\leavevmode|| |^|\raise||.16ex\hbox{#1}}|
|\font\smallheadfont=cmr8 |^|at|| 8truept|
|\font\largeheadfont=cmdunh10 at 14.4truept|
|\font\logofont=manfnt at 14.4truept|
\smallbreak
|\def\rjdletterhead{|
|  \def\sendingaddress{R. J. DROFNATS, F.T.U.G.\par|^^{TeX Users Group}
|    PROFESSOR OF FARM ECOLOGY\par|
|    TEX.RJD @ SU-SCORE.ARPA\par|^^{atsign}
|    \up[415\up]\thinspace 497-4975\par}|
|  \def\returnaddress{R. J. Drofnats, Dept.~of Farm Ecology\par|
|    The University of St.~Anford\par|
|    P. O. Box 1009, Haga Alto, CA 94321 USA}|
|  \letterhead}|
\smallbreak
|\def\letterhead{\pageno=1 \def\addressee{} \univletterhead|
|  {\leftskip=\longindentation|
|    {\baselineskip9truept\smallheadfont\sendingaddress}|
|    \bigskip\bigskip\rm\today\bigskip}}|
\smallbreak
|\def\univletterhead{\vglue-\voffset|
|  \hbox{\hbox to\longindentation{\raise4truemm\hbox{\logofont|
|        \kern2truept X\kern-1.667truept|
|        \lower2truept\hbox{X}\kern-1.667truept X}\hfil|
|      \largeheadfont The University of St.~Anford\hfil}%|
|    \kern-\longindentation|
|    \vbox{\smallheadfont\baselineskip9truept|
|      \leftskip=\longindentation BOX 1009\par HAGA ALTO, CA 94321}}|
|  \vskip2truept\hrule\vskip4truept }|
\smallbreak
|\def|^|\makelabel||{\endletter\hbox{\vrule|
|    \vbox{\hrule \kern6truept|
|      \hbox{\kern6truept\vbox to 2truein{\hsize=\longindentation|
|          \smallheadfont\baselineskip9truept\returnaddress|
|          \vfill\moveright 2truein\copy\theaddress\vfill}%|
|        \kern6truept}\kern6truept\hrule}\vrule}|
|  \pageno=0\vfill\eject}|
\endlines

\medbreak
Our second example is a format for ^{concert programs}, to be used in
^^{music} ^^{programs, for music}
connection with orchestra performances, recitals, and the like. We shall
assume that the entire program fits on a single page, and that the
copy is to be 4~inches wide. Comparatively large type ($12\pt$) will
normally be used, but there is a provision for $10\pt$ and even ^^{sizes
of type} $8\pt$ type in case the program includes pieces with a lot of
subparts (e.g., ^{Bach}'s Mass in B~minor, or ^{Beethoven}'s ^{Diabelli}
Variations). To select the ^{type size}, a user says ^|\bigtype|,
^|\medtype|, or ^|\smalltype|, respectively. These macros for ^{size switching}
are comparatively simple because concert programs don't require any
mathematics; hence the math fonts don't need to be changed. On the other
hand, the format does take sharp and flat signs from the ``^{math italic}''
font, which it calls `|\mus|':
\beginlines
|\font\twelverm=cmr12|
|\font\twelvebf=cmbx12|
|\font\twelveit=cmti12|
|\font\twelvesl=cmsl12|
|\font\twelvemus=cmmi12|
\smallskip
|\font\eightrm=cmr8|
|\font\eightbf=cmbx8|
|\font\eightit=cmti8|
|\font\eightsl=cmsl8|
|\font\eightmus=cmmi8|
\smallskip
|\def\bigtype{\let|^|\rm||=\twelverm \let|^|\bf||=\twelvebf|
|  \let|^|\it||=\twelveit \let|^|\sl||=\twelvesl \let\mus=\twelvemus|
|  \baselineskip=14pt minus 1pt|
|  \rm}|
|\def\medtype{\let\rm=\tenrm \let\bf=\tenbf|
|  \let\it=\tenit \let\sl=\tensl \let\mus=\teni|
|  \baselineskip=12pt minus 1pt|
|  \rm}|
|\def\smalltype{\let\rm=\eightrm \let\bf=\eightbf|
|  \let\it=\eightit \let\sl=\eightsl \let\mus=\eightmus|
|  \baselineskip=9.5pt minus .75pt|
|  \rm}|
\smallskip
|\hsize=4in|
|\nopagenumbers|
|\bigtype|
\endlines
Notice the ^{shrinkability} in the ^|\baselineskip| settings. This would be
undesirable in a book format, because different ^{spacing between lines}
^^{leading} on different pages would look bad; but in a one-page document
it helps squeeze the copy to fit the page, in an emergency. \ (There's no
need for ^{stretchability} in the baselineskip here, because a |\vfill| will
be used at the bottom of the page.)

Musical programs have a specialized vocabulary, and it is desirable to define
a few control sequences for things that plain \TeX\ doesn't make as convenient
as they could be for this particular application:
\beginlines
|\def\(#1){{\rm(}#1\/{\rm)}}|
|\def\sharp{\raise.4ex\hbox{\mus\char"5D}}|
|\def\flat{\raise.2ex\hbox{\mus\char"5B}}|
|\let\,=|^|\thinspace|
\endlines
The ^|\(| macro produces roman ^{parentheses} in the midst of ^{italicized
text}; the ^|\sharp| and ^|\flat| macros produce musical signs in the
current type size. The ^|\,|~macro makes it easy to specify
the ^{thin space} that is used in constructions like
`K.\thinspace550' ^^{K\"ochel}^^{Mozart} % his 40th symphony
and `Op.\thinspace59'. ^^{Dvo\v r\'ak} % his Legends
\ (Plain \TeX\ has already defined |\,| and |\sharp| and |\flat| in a different
way; but those definitions apply only to math formulas, so they aren't
relevant in this application.)

Before discussing the rest of the music macros, let's take a look at a complete
example. The next two pages show the input and output for a typical concert
program.

\eject
\begingroup \parindent=0pt \obeylines
|\input concert|
\medskip
|\tsaologo|
|\medskip|
|\centerline{Friday, November 19, 1982, 8:00 p.m.}|
|\bigskip|
|\centerline{\bf PROGRAM}|
|\medskip|
\medskip
|\composition{Variations on a Theme by Tchaikovsky}|%
  ^^{Tchaikovsky, see Cha\u\i...}
|\composer{Anton S. Arensky (1861--1906)}|^^{Arenski\u\i}
|\smallskip|
|{\medtype|
|\movements{Tema: Moderato\cr|
|  Var.~I: Un poco pi\`u mosso&Var.~V: Andante\cr|
|  Var.~II: Allegro non troppo&Var.~VI: Allegro con spirito\cr|
|  Var.~III: Andantino tranquillo&Var.~VII: Andante con moto\cr|
|  Var.~IV: Vivace&Coda: Moderato\cr}|
|}|
\medskip
|\bigskip|
\medskip
|\composition{Concerto for Horn and Hardart, S.\,27}|
|\composer{P. D. Q. Bach (1807--1742)?}|^^{Bach, PDQ}
|\smallskip|
|\movements{Allegro con brillo\cr|
|  Tema con variazione \(su un tema differente)\cr|
|  Menuetto con panna e zucchero\cr}|
|\medskip|
|\soloists{Ben Lee User, horn\cr|^^{User}
|  Peter Schickele, hardart\cr}|^^{Schickele}
\medskip
|\bigskip|
|\centerline{INTERMISSION}|
|\bigskip|
\medskip
|\composition{Symphony No.\,3 in E\flat\ Major\cr|
|  Op.\,55, ``The Eroica''\cr}|
|\composer{Ludwig van Beethoven (1770--1827)}|^^{Beethoven}
|\smallskip|
|\movements{Allegro con brio\cr|
|  Marcia funebre: Adagio assai\cr|
|  Scherzo: Allegro vivace\cr|
|  Finale: Allegro molto\cr}|
\medskip
|\bigskip|
|\smalltype \noindent|
|Members of the audience are kindly requested to turn off the|
|alarms on their digital watches, and to cough only between movements.|
\medskip
|\bye|
\eject\endgroup
\hbox to\hsize{\hss \vbox{ % \centerline is not a \long macro!
\font\twelverm=cmr12
\font\twelvebf=cmbx12
\font\twelveit=cmti12
\font\twelvesl=cmsl12
\font\twelvemus=cmmi12

\def\bigtype{\let\rm=\twelverm \let\bf=\twelvebf \let\it=\twelveit
  \let\sl=\twelvesl \let\mus=\twelvemus \baselineskip=14pt minus 1pt \rm}
\def\medtype{\let\rm=\tenrm \let\bf=\tenbf \let\it=\tenit
  \let\sl=\tensl \let\mus=\teni \baselineskip=12pt minus 1pt \rm}
\def\smalltype{\let\rm=\eightrm \let\bf=\eightbf \let\it=\eightit
  \let\sl=\eightsl \let\mus=\eighti \baselineskip=9.5pt minus .75pt \rm}

\def\flat{\raise.2ex\hbox{\mus\char"5B}}
\def\sharp{\raise.4ex\hbox{\mus\char"5D}}
\let\,=\thinspace
\def\(#1){{\rm(}#1\/{\rm)}}

\hsize=4in
\bigtype

\def\composition#1{\halign{\bf\quad##\hfil\cr
    \kern-1em#1\crcr}} % use \cr's if more than one line
\def\composer#1{\rightline{\bf#1}}
\def\movements#1{\halign{\quad\it##\hfil&&\qquad\it##\hfil\cr#1\crcr}}
\def\soloists#1{\centerline{\bf\vbox{\halign{##\hfil\cr#1\crcr}}}}

\def\tsaologo{\vbox{\bigtype\bf
    \line{\hrulefill}
    \kern-.5\baselineskip
    \line{\hrulefill\phantom{ THE ST.\,ANFORD ORCHESTRA }\hrulefill}
    \kern-.5\baselineskip
    \line{\hrulefill\hbox{ THE ST.\,ANFORD ORCHESTRA }\hrulefill}
    \kern-.5\baselineskip
    \line{\hrulefill\phantom{ R. J. Drofnats, Conductor }\hrulefill}
    \kern-.5\baselineskip
    \line{\hrulefill\hbox{ R. J. Drofnats, Conductor }\hrulefill}
    }}
\tsaologo
\medskip
\centerline{Friday, November 19, 1982, 8:00 p.m.}
\bigskip
\centerline{\bf PROGRAM}
\medskip

\composition{Variations on a Theme by Tchaikovsky}
\composer{Anton S. Arensky (1861--1906)}
\smallskip
{\medtype
\movements{Tema: Moderato\cr
  Var.~I: Un poco pi\`u mosso&Var.~V: Andante\cr
  Var.~II: Allegro non troppo&Var.~VI: Allegro con spirito\cr
  Var.~III: Andantino tranquillo&Var.~VII: Andante con moto\cr
  Var.~IV: Vivace&Coda: Moderato\cr}
}

\bigskip

\composition{Concerto for Horn and Hardart, S.\,27}
\composer{P. D. Q. Bach (1807--1742)?}
\smallskip
\movements{Allegro con brillo\cr
  Tema con variazione \(su un tema differente)\cr
  Menuetto con panna e zucchero\cr}
\medskip
\soloists{Ben Lee User, horn\cr
  Peter Schickele, hardart\cr}

\bigskip
\centerline{INTERMISSION}
\bigskip

\composition{Symphony No.\,3 in E\flat\ Major\cr
  Op.\,55, ``The Eroica''\cr}
\composer{Ludwig van Beethoven (1770--1827)}
\smallskip
\movements{Allegro con brio\cr
  Marcia funebre: Adagio assai\cr
  Scherzo: Allegro vivace\cr
  Finale: Allegro molto\cr}

\bigskip
\smalltype \noindent
Members of the audience are kindly requested to turn off the
alarms on their digital watches, and to cough only between movements.
}\hss}^^{Drofnats}\vfill\eject

\edef\\{\hskip\the\parindent} % I'm emphasizing this "design element" here!
Most of the macros in |concert.tex| have already been defined. Plain \TeX\
takes care of things like |\centerline| and |\bigskip|, so only
|\composition|, |\composer|, |\movements|, and |\soloists| remain to
be specified:
\beginlines
|\def\composition#1{\halign{\bf\quad##\hfil\cr|
\\|\kern-1em#1\crcr}} % use \cr's if more than one line|
|\def\composer#1{\rightline{\bf#1}}|
|\def\movements#1{\halign{\quad\it##\hfil&&\qquad\it##\hfil\cr#1\crcr}}|
|\def\soloists#1{\centerline{\bf\vbox{\halign{##\hfil\cr#1\crcr}}}}|
\endlines
The |\composition| macro is set up to put the title of the composition on
two or more lines, if needed, but a single line usually suffices. Notice
that ^|\crcr| has been used so that the final ^|\cr| in the argument to
|\composition| is not needed. Similarly, |\movements| might be used to
produce only a single line, and |\soloists| might be used when there
is only one soloist.

There's also a |\tsaologo| macro. It applies only to one particular
orchestra, but the definition is somewhat interesting nonetheless:
\beginlines
|\def\tsaologo{\vbox{\bigtype\bf|
\\^|\line||{|^|\hrulefill||}|
\\|\kern-.5\baselineskip|
\\|\line{\hrulefill|^|\phantom||{ THE ST.\,ANFORD ORCHESTRA }\hrulefill}|
\\|\kern-.5\baselineskip|
\\|\line{\hrulefill\hbox{ THE ST.\,ANFORD ORCHESTRA }\hrulefill}|
\\|\kern-.5\baselineskip|
\\|\line{\hrulefill\phantom{ R. J. Drofnats, Conductor }\hrulefill}|
\\|\kern-.5\baselineskip|
\\|\line{\hrulefill\hbox{ R. J. Drofnats, Conductor }\hrulefill}|
\\|}}|
\endlines

The author ^^{Knuth} has extended these macros to a more elaborate format
that includes special features for listing the members of the orchestra and
for program notes, etc.; in this way it becomes fairly easy to typeset little
booklets for concert patrons. Such extensions need not be discussed further
in this appendix, because they don't illustrate any essentially new ideas.

Notice that the |\composition| and |\movements| and |\soloists| macros do
not include any special provision for vertical ^{spacing}; the user is
supposed to insert ^|\smallskip|, ^|\medskip|, and ^|\bigskip| as
desired. This was done deliberately, because different concert programs
seem to demand different spacing; no automatic scheme actually works very
well in practice, since musical literature is so varied.

\medbreak
Let's turn now to the design of a format for an entire book, using this book
itself as an example. How did the author prepare the computer file
that generated {\sl The \TeX book\/}?  We have already seen several hundred
pages of output produced from that file; our goal in the remainder of this
appendix will be to examine the input that was used behind the scenes.

In the first place, the author prepared ^{sample pages} and showed them to
the publisher's book designer. \ (The importance of this step cannot be
overemphasized. There is a danger that authors---who are now able to
typeset their own books with \TeX---will attempt to do their own designs,
^^{author, typesetting by}
without professional help. ^{Book design} is an art that requires considerable
creativity, skill, experience, and taste; it is one of the most important
services that a publisher traditionally provides to an author.)

Sample pages that are used as the basis of a design should show each of
the elements in the book. In this case the elements included chapter titles,
illustrations, subchapter headings, footnotes, displayed formulas,
typewriter type, dangerous bends, exercises, answers, quotations,
tables, numbered lists, bulleted lists, etc.; the author also expressed
a desire for generous margins, so that readers could make marginal notes.

The designer, Herb ^{Caswell}, faced a difficult problem of bringing all
those disparate elements into a consistent framework. He decided to achieve
this by using a uniform indentation of 3~^{picas} for normal paragraph openings
as well as for dangerous bends; and to establish this element of the design
by using it also for all the displayed material, instead of centering the
displays.

He decided to put the page numbers in bold type, out in the margins (where
there was plenty of room, thanks to the author's request for white space);
and he decided to use italic type with caps and lower case for the running
headlines, so that the pages would have a somewhat informal flavor.

He chose 10-point type (on a 12-point base) for the main text, and 9-point type
(on an 11-point base) for the dangerous bends; the typeface was predetermined.
He~chose an ^|\hsize| of 29~picas and a ^|\vsize| of 44~picas. He decided
to give subheadings like `\kern1pt{\manual\char'170}\kern.15em
{\eightbf EXERCISE \bf13.8}' in boldface caps before the statement of each
exercise. He~specified the amount of vertical space before and after such
things as exercises, dangerous-bend paragraphs, and displayed equations.
He decided to devote an entire left-hand page to each chapter illustration.
And so on; each decision influenced the others, so that the final book would
appear to be as coherent and attractive as possible under the circumstances.
After the main portion of the book was designed, he worked out a format for
the ^{front matter} (i.e., the pages that precede page~1); he arranged to
have the same amount of ``^{sinkage}'' (white space) at the top of each page
there, so that the opening pages of the book would look unified and ``open.''

The author hasn't actually followed the designer's specifications in every
detail. For example, nothing about stretching or shrinking of vertical spaces
appeared in the design specs; the author introduced the notion of flexible
glue on his own initiative, based on his observations of cut-and-paste
operations often used in page makeup. If this book has any beauties, they
should be ascribed to Herb Caswell; if it has any blemishes, they should be
ascribed to Don Knuth, who wrote the formatting
macros that we are now about to discuss.

The computer file |texbook.tex| that generated {\sl The \TeX book\/} begins
with a copyright notice, and then it says `|\input| |manmac|'. The auxiliary
file ^|manmac.tex| contains the formatting macros, and it begins by
loading 9-point, 8-point, and 6-point~fonts:
\beginlines
|\font\ninerm=cmr9   \font\eightrm=cmr8   \font\sixrm=cmr6|
|\font\ninei=cmmi9   \font\eighti=cmmi8   \font\sixi=cmmi6|
|\font\ninesy=cmsy9  \font\eightsy=cmsy8  \font\sixsy=cmsy6|
|\font\ninebf=cmbx9  \font\eightbf=cmbx8  \font\sixbf=cmbx6|
|\font\ninett=cmtt9  \font\eighttt=cmtt8|
|\font\nineit=cmti9  \font\eightit=cmti8|
|\font\ninesl=cmsl9  \font\eightsl=cmsl8|
\endlines
(These fonts had been ^|\preloaded| in Appendix B\null; now they're officially
loaded.)

The fonts intended for math formulas need to have a nonstandard ^|\skewchar|.
The typewriter fonts are given ^|\hyphenchar||=-1| so that ^{hyphenation}
is inhibited when control sequence names and keywords appear in the text
of a paragraph.
\beginlines
|\skewchar\ninei='177  \skewchar\eighti='177  \skewchar\sixi='177|
|\skewchar\ninesy='60  \skewchar\eightsy='60  \skewchar\sixsy='60|
|\hyphenchar\ninett=-1 \hyphenchar\eighttt=-1 \hyphenchar\tentt=-1|
\endlines

A few more fonts are needed for special purposes:
\beginlines
|\font\tentex=cmtex10               % TeX character set as in Appendix C|
|\font\inchhigh=cminch              % inch-high caps for chapter openings|
|\font\titlefont=cmssdc10 at 40pt   % titles in chapter openings|
|\font\eightss=cmssq8               % quotations in chapter closings|
|\font\eightssi=cmssqi8             % ditto, slanted|
|\font\tenu=cmu10                   % unslanted text italic|
|\font\manual=manfnt                % METAFONT logo and special symbols|
|\font\magnifiedfiverm=cmr5 at 10pt % to demonstrate magnification|
\endlines

Now we come to the ^{size-switching} macros, which are much more
elaborate than they were in the previous example because mathematics
needs to be supported in three different sizes. The format also
provides for a pseudo ``^{small caps}'' (|\sc|); a true caps-and-small-caps
font was not really necessary in the few cases that |\sc| was used.
A~dimension variable ^|\ttglue| is set equal to the desired spacing for
the typewriter-like text that occasionally appears in paragraphs; the
|\tt| fonts have fixed spacing, which doesn't mix well with
variable spacing, hence the macros below use |\ttglue| between words in
appropriate places.
\beginlines
|\catcode`@=11 % we will access private macros of plain TeX (carefully)|
^|\newskip||\ttglue|^^{atsign}
\smallbreak
|\def|^|\tenpoint||{\def|^|\rm||{|^|\fam||0\tenrm}% switch to 10-point type|
|  |^|\textfont||0=\tenrm \scriptfont0=\sevenrm \scriptscriptfont0=\fiverm|
|  \textfont1=\teni  |^|\scriptfont||1=\seveni  \scriptscriptfont1=\fivei|
|  \textfont2=\tensy \scriptfont2=\sevensy |^|\scriptscriptfont||2=\fivesy|
|  \textfont3=\tenex \scriptfont3=\tenex   \scriptscriptfont3=\tenex|
|  \textfont|^|\itfam||=\tenit  \def|^|\it||{\fam\itfam\tenit}%|
|  \textfont|^|\slfam||=\tensl  \def|^|\sl||{\fam\slfam\tensl}%|
|  \textfont|^|\ttfam||=\tentt  \def|^|\tt||{\fam\ttfam\tentt}%|
|  \textfont|^|\bffam||=\tenbf  \scriptfont\bffam=\sevenbf|
|   \scriptscriptfont\bffam=\fivebf  \def|^|\bf||{\fam\bffam\tenbf}%|
|  \tt \ttglue=.5em plus.25em minus.15em|
|  |^|\normalbaselineskip=12pt|
|  \setbox|^|\strutbox||=\hbox{\vrule height8.5pt depth3.5pt width0pt}%|
|  \let|^|\sc||=\eightrm  \let|^|\big||=\tenbig  \normalbaselines\rm}|
\smallbreak
|\def|^|\ninepoint||{\def\rm{\fam0\ninerm}% switch to 9-point type|
|  \textfont0=\ninerm \scriptfont0=\sixrm \scriptscriptfont0=\fiverm|
|  \textfont1=\ninei  \scriptfont1=\sixi  \scriptscriptfont1=\fivei|
|  \textfont2=\ninesy \scriptfont2=\sixsy \scriptscriptfont2=\fivesy|
|  \textfont3=\tenex  \scriptfont3=\tenex \scriptscriptfont3=\tenex|
|  \textfont\itfam=\nineit  \def\it{\fam\itfam\nineit}%|
|  \textfont\slfam=\ninesl  \def\sl{\fam\slfam\ninesl}%|
|  \textfont\ttfam=\ninett  \def\tt{\fam\ttfam\ninett}%|
|  \textfont\bffam=\ninebf  \scriptfont\bffam=\sixbf|
|   \scriptscriptfont\bffam=\fivebf  \def\bf{\fam\bffam\ninebf}%|
|  \tt \ttglue=.5em plus.25em minus.15em|
|  \normalbaselineskip=11pt|
|  \setbox\strutbox=\hbox{\vrule height8pt depth3pt width0pt}%|
|  \let\sc=\sevenrm  \let\big=\ninebig  \normalbaselines\rm}|
\smallbreak
|\def|^|\eightpoint||{\def\rm{\fam0\eightrm}% switch to 8-point type|
|  \textfont0=\eightrm \scriptfont0=\sixrm \scriptscriptfont0=\fiverm|
|  \textfont1=\eighti  \scriptfont1=\sixi  \scriptscriptfont1=\fivei|
|  \textfont2=\eightsy \scriptfont2=\sixsy \scriptscriptfont2=\fivesy|
|  \textfont3=\tenex   \scriptfont3=\tenex \scriptscriptfont3=\tenex|
|  \textfont\itfam=\eightit  \def\it{\fam\itfam\eightit}%|
|  \textfont\slfam=\eightsl  \def\sl{\fam\slfam\eightsl}%|
|  \textfont\ttfam=\eighttt  \def\tt{\fam\ttfam\eighttt}%|
|  \textfont\bffam=\eightbf  \scriptfont\bffam=\sixbf|
|   \scriptscriptfont\bffam=\fivebf  \def\bf{\fam\bffam\eightbf}%|
|  \tt \ttglue=.5em plus.25em minus.15em|
|  \normalbaselineskip=9pt|
|  \setbox\strutbox=\hbox{\vrule height7pt depth2pt width0pt}%|
|  \let\sc=\sixrm  \let\big=\eightbig  \normalbaselines\rm}|
\smallbreak
|\def\tenbig#1{{\hbox{$\left#1\vbox to8.5pt{}\right.\n@space$}}}|
|\def\ninebig#1{{\hbox{$\textfont0=\tenrm\textfont2=\tensy|
|  \left#1\vbox to7.25pt{}\right.\n@space$}}}|
|\def\eightbig#1{{\hbox{$\textfont0=\ninerm\textfont2=\ninesy|
|  \left#1\vbox to6.5pt{}\right.\n@space$}}}|
|\def\tenmath{\tenpoint\fam-1 } % for 10-point math in 9-point territory|
\endlines

Issues of page layout are dealt with next. First, the basics:^^|pc|
\beginlines
^|\newdimen||\pagewidth \newdimen\pageheight \newdimen\ruleht|
^|\hsize||=29pc  |^|\vsize||=44pc  |^|\maxdepth||=2.2pt  |^|\parindent||=3pc|
|\pagewidth=\hsize \pageheight=\vsize \ruleht=.5pt|
^|\abovedisplayskip||=6pt plus 3pt minus 1pt|
^|\belowdisplayskip||=6pt plus 3pt minus 1pt|
^|\abovedisplayshortskip||=0pt plus 3pt|
^|\belowdisplayshortskip||=4pt plus 3pt|
\endlines
(The curious value of\/ |\maxdepth| was chosen only to provide an example
in Chapter~15; there's no deep reason behind it.)

When the author prepared this book, he made notes about what things ought
to go into the index from each page. These notes were shown in small type
on his proofsheets, like the words `^{marginal hacks}' in the right
\insert\margin{\hbox{\marginstyle mar-}}%
\insert\margin{\hbox{\marginstyle ginal}}%
\insert\margin{\hbox{\marginstyle hacks}}%
margin of this page. The |manmac| format uses an insertion class called
^|\margin| to handle such notes.
\beginlines
^|\newinsert||\margin|
|\dimen\margin=\maxdimen % no limit on the number of marginal notes|
|\count\margin=0 \skip\margin=0pt % marginal inserts take up no space|
\endlines

The ^|\footnote| macro of plain \TeX\ needs to be changed because
^{footnotes} are indented and set in 8-point type. Some simplifications have
also been made, since footnotes are used so infrequently in this book.
\beginlines
|\def\footnote#1{\edef\@sf{\spacefactor\the\spacefactor}#1\@sf|
|      \insert\footins\bgroup\eightpoint|
|      \interlinepenalty100 \let\par=|^|\endgraf|
|        \leftskip=0pt \rightskip=0pt|
|        \splittopskip=10pt plus 1pt minus 1pt \floatingpenalty=20000|
|        \smallskip|^|\item||{#1}\bgroup\strut\aftergroup\@foot\let\next}|
|\skip\footins=12pt plus 2pt minus 4pt % space added when footnote exists|
|\dimen\footins=30pc % maximum footnotes per page|
\endlines

The text of ^{running headlines} will be kept in a control sequence
called |\rhead|. Some pages should not have headlines; the |\titlepage|
macro suppresses the headline on the next page that is output.
\beginlines
^|\newif||\iftitle|
|\def\titlepage{\global\titletrue} % for pages without headlines|
|\def\rhead{} % \rhead contains the running headline|
\smallskip
|\def\leftheadline{\hbox to \pagewidth{|^|\spaceskip||=0pt|
|    \vbox to 10pt{}% strut to position the baseline|^^{strut}
|    |^|\llap||{\tenbf|^|\folio|^|\kern||1pc}% folio to left of text|
|    \tenit\rhead\hfil}} % running head flush left|
|\def\rightheadline{\hbox to \pagewidth{\spaceskip=0pt\vbox to 10pt{}%|
|    \hfil\tenit\rhead\/% running head flush right|
|    |^|\rlap||{\kern1pc\tenbf\folio}}} % folio to right of text|
\endlines

Pages are shipped to the output by the |\onepageout| macro, which
attaches headlines, marginal notes, and/or footnotes, as appropriate.
Special ^{registration marks} are typeset at the top of title pages, so
that the pages will line up properly on printing plates that are made
photographically from \TeX's ``camera-ready'' output. ^^{camera alignment}
A small page number is also printed next to the corner markings; such
auxiliary information will, of course, be erased before the pages are
actually printed.
\beginlines
|\def\onepageout#1{\shipout\vbox{ % here we define one page of output|
|    |^|\offinterlineskip|| % butt the boxes together|
|    \vbox to 3pc{ % this part goes on top of the 44pc pages|
|      \iftitle \global\titlefalse \setcornerrules|
|      \else|^|\ifodd||\pageno\rightheadline\else\leftheadline\fi\fi \vfill}|
|    \vbox to \pageheight{|
|      |^|\ifvoid||\margin\else % marginal info is present|
|       |^|\rlap||{\kern31pc\vbox to0pt{\kern4pt\box\margin \vss}}\fi|
|      #1 % now insert the main information|
|      \ifvoid\footins\else % footnote info is present|
|       \vskip\skip\footins \kern-3pt|
|       \hrule height\ruleht width\pagewidth \kern-\ruleht \kern3pt|
|       \unvbox\footins\fi|
|      \boxmaxdepth=\maxdepth}}|
|  |^|\advancepageno||}|
\smallbreak
|\def\setcornerrules{\hbox to \pagewidth{% for camera alignment|
|    \vrule width 1pc height\ruleht \hfil \vrule width 1pc}|
|  \hbox to \pagewidth{|^|\llap||{\sevenrm(page \folio)\kern1pc}%|
|    \vrule height1pc width\ruleht depth0pt|
|    \hfil \vrule width\ruleht depth0pt}}|
\smallbreak
^|\output||{\onepageout{|^|\unvbox||255}}|
\endlines

A different ^{output routine} is needed for Appendix~I (the index), because
most of that appendix appears in ^{two-column format}. Instead of handling
double columns with an `|\lr|' switch, as discussed in Chapter~23,
|manmac| does the job with ^|\vsplit|, after collecting more than
enough material to fill a page. This approach makes it comparatively
easy to ^{balance the columns} on the last page of the index. A~more
difficult approach would be necessary if the index contained
insertions (e.g., footnotes); fortunately, it doesn't.
Furthermore, there is no need to use ^|\mark| as suggested in the
index example of Chapter~23, since the entries in Appendix~I tend to
be quite short. The only real complication that |manmac| faces
is the fact that Appendix~I begins and ends with single-column format;
partial pages need to be juggled carefully as the format changes back and forth.
\beginlines
^|\newbox||\partialpage|
|\def\begindoublecolumns{\begingroup|
|  \output={\global\setbox\partialpage=\vbox{\unvbox255\bigskip}}\eject|
|  \output={\doublecolumnout} |^|\hsize||=14pc |^|\vsize||=89pc}|
|\def\enddoublecolumns{\output={\balancecolumns}\eject|
|  \endgroup \pagegoal=\vsize}|
\smallbreak
|\def\doublecolumnout{\splittopskip=\topskip |^|\splitmaxdepth||=\maxdepth|
|  \dimen@=44pc \advance\dimen@ by-\ht\partialpage|
|  \setbox0=\vsplit255 to\dimen@ \setbox2=\vsplit255 to\dimen@|
|  \onepageout\pagesofar \unvbox255 \penalty|^|\outputpenalty||}|
|\def\pagesofar{\unvbox\partialpage|
|  |^|\wd||0=\hsize \wd2=\hsize \hbox to\pagewidth{\box0\hfil\box2}}|
|\def\balancecolumns{\setbox0=\vbox{\unvbox255} \dimen@=\ht0|
|  \advance\dimen@ by\topskip \advance\dimen@ by-\baselineskip|
|  |^|\divide||\dimen@ by2 |^|\splittopskip||=\topskip|
|  {\vbadness=10000 |^|\loop|| \global\setbox3=\copy0|
|    \global\setbox1=\vsplit3 to\dimen@|
|    |^|\ifdim|^|\ht||3>\dimen@ \global\advance\dimen@ by1pt \repeat}|
|  \setbox0=\vbox to\dimen@{\unvbox1} \setbox2=\vbox to\dimen@{\unvbox3}|
|  \pagesofar}|
\endlines
The balancing act sets ^|\vbadness| infinite while it is searching for
a suitable column height, so that ^{underfull} vboxes won't be reported
unless the actual columns are bad after balancing. The columns in Appendix~I
have a lot of stretchability, since there's a ^|\parskip|
of |0pt| |plus|~|.8pt| between adjacent entries, and since there is room for
more than 50 lines per column; therefore the |manmac| balancing routine tries
to make both the top and bottom baselines agree at the end of the index.
In applications where the glue is not so flexible it would be more
appropriate to let the right-hand column be a little short; the best
way to do this is probably to replace the command `|\unvbox3|' by
`|\dimen2=|^|\dp||3| |\unvbox3| |\kern-\dimen2| ^|\vfil|'.

The next macros are concerned with chapter formatting. Each chapter in the
manuscript file starts out with the macro ^|\beginchapter|; it ends
with ^|\endchapter| and two ^{quotations}, ^^{epigraphs} followed
by ^|\eject|. For example, Chapter~15 was generated by \TeX\ commands
that look like this in the file |texbook.tex|:
\beginlines
|\beginchapter Chapter 15. How \TeX\ Makes\\Lines into Pages|
\medskip
|\TeX\ attempts to choose desirable places to divide your document into|
\qquad\dots\quad (about 1100 lines of the manuscript are omitted here)
|break. \ (Deep breath.) \ You got that?|
\medskip
|\endchapter|
\medskip
|Since it is impossible to foresee how [footnotes] will happen to come out|
|in the make-up, it is |%
  |impracticable to number them from 1 up on each page.|\kern-1em
|The best way is to number them consecutively throughout an article|
|or by chapters in a book.|
|\author UNIVERSITY OF CHICAGO PRESS, {\sl Manual of Style\/} (1910)|
\medskip^^{CHICAGO}^^{KNUTH}
|\bigskip|
\medskip
|Don't use footnotes in your books, Don.|
|\author JILL KNUTH (1962)|
\medskip
|\eject|
\endlines
The `|\\|' in the title line specifies a line break to be used on the
left-hand title page that faces the beginning of the chapter. Most of the
|\beginchapter| macro is devoted to preparing that title page; the
^|\TeX| logo needs somewhat different spacing when it is typeset in
|\titlefont|, and the |\inchhigh| digits need to be brought closer
together in order to look right in a title.
\beginlines
^|\newcount||\exno % for the number of exercises in the current chapter|
|\newcount\subsecno % for the number of subsections in the current chapter|
\smallbreak
^|\outer||\def\beginchapter#1 #2#3. #4\par{\def\chapno{#2#3}|
|  \global\exno=0 \subsecno=0|
|  \ifodd\pageno|
|   |^|\errmessage||{You had too much text on that last page; I'm backing up}|
|   \advance\pageno by-1 \fi|
|  \def\\{ } % \\'s in the title will be treated as spaces|
|  |^|\message||{#1 #2#3:} % show the chapter title on the terminal|
|  |^|\xdef||\rhead{#1 #2#3: #4\unskip} % establish a new running headline|
|  {\def\TeX{T\kern-.2em\lower.5ex\hbox{E}\kern-.06em X}|
|    \def\\{#3} \advance\hsize by -18mm|
|    |^|\ifx||\empty\\ \rightline{\inchhigh #2\kern-.04em}|
|    \else\rightline{\inchhigh #2\kern-.06em#3\kern-.04em}\fi|
|    \vskip1.75pc \baselineskip=36pt \lineskiplimit=1pt \lineskip=12pt|
|    \let\\=|^|\cr|| % now the \\'s are line dividers|
|    \halign{\line{\titlefont\hss##}\\#4\unskip\\}|
|    \titlepage\vfill\eject} % output the chapter title page|
|  \tenpoint\noindent\ignorespaces} % the first paragraph is not indented|
\endlines

An extra page is ejected at the end of a chapter, if necessary, so
that the closing quotations will occur on a right-hand page. \ (The
logic for doing this is not perfect, but it doesn't need to be, because
it fails only when the chapter has to be shortened or lengthened anyway;
book preparation with \TeX, as with type, encourages interaction between
humans and machines.) \
The lines of the quotations are set ^{flush right} by using
^|\obeylines| together with a stretchable ^|\leftskip|:
\beginlines
|\outer\def\endchapter{\ifodd\pageno \else\vfill|^|\eject||\null\fi|
|  \begingroup\bigskip\vfill % beginning of the quotes|
|  \def\eject{|^|\endgroup||\eject} % ending of the quotes|
|  \def\par{\ifhmode\/\endgraf\fi}\obeylines|
|  \def|^|\TeX||{T\kern-.2em\lower.5ex\hbox{E}X}|
|  \eightpoint \let\tt=\ninett \baselineskip=10pt |^|\interlinepenalty||=10000|
|  \leftskip=0pt plus 40pc minus \parindent |^|\parfillskip||=0pt|
|  \let|^|\rm||=\eightss \let|^|\sl||=\eightssi \everypar{\sl}}|
|\def\author#1(#2){\smallskip\noindent\rm--- #1|^|\unskip|^|\enspace||(#2)}|
\endlines

We come now to what goes on inside the chapters themselves. Dangerous and
doubly dangerous bends are specified by typing `^|\danger|' or `^|\ddanger|'
just before a paragraph that is supposed to display a warning symbol:
\beginlines
|\def\dbend{{\manual\char127}} % "dangerous bend" sign|^^{dangerous bend}
|\def\d@nger{\medbreak|^|\begingroup|^|\clubpenalty||=10000|
|  \def\par{|^|\endgraf|^|\endgroup|^|\medbreak||} |%
  ^|\noindent|^|\hang|^|\hangafter||=-2|
|  \hbox to0pt{\hskip-\hangindent\dbend\hfill}|^|\ninepoint||}|
^|\outer||\def\danger{\d@nger}|
|\def\dd@nger{\medbreak\begingroup\clubpenalty=10000|
|  \def\par{\endgraf\endgroup\medbreak} \noindent\hang\hangafter=-2|
|  \hbox to0pt{\hskip-\hangindent\dbend\kern1pt\dbend\hfill}\ninepoint}|
|\outer\def\ddanger{\dd@nger}|
|\def\enddanger{\endgraf\endgroup} % omits the \medbreak|
\endlines
(It's necessary to type `|\enddanger|' at the end of a dangerous bend
only in rare cases that a medium space is not desired after the paragraph;
e.g., `|\smallskip|^|\item|' might be used to give an itemized list within
the scope of the dangerous bend sign.)

A few chapters and appendices of this book (e.g., Chapter~18 and Appendix~B)
are divided into numbered subsections. Such subsections are specified
in the manuscript by typing, for example,
\begintt
\subsection Allocation of registers.
\endtt
Appendix A is subdivided in another way, by paragraphs that have answer numbers:
\beginlines
|\outer\def\subsection#1. {\medbreak\advance\subsecno by 1|
|  \noindent{\it \the\subsecno.\enspace#1.\enspace}}|
|\def\ansno#1.#2:{\medbreak\noindent|
|  \hbox to\parindent{\bf\hss#1.#2.\enspace}\ignorespaces}|
\endlines
We will see below that the manuscript doesn't actually specify an |\ansno|
directly; each call of\/ |\ansno| is generated automatically by the
|\answer| macro.

Appendix H points out {\sl The \TeX book\/} calls for
three hyphenation exceptions:
\beginlines
^|\hyphenation||{man-u-script man-u-scripts ap-pen-dix}|
\endlines

A few macros in |manmac| provide special constructions that are occasionally
needed in paragraphs: |\MF| for ^^{METAFONT}`\MF', |\AmSTeX| for ^^{AmSTeX}
`\AmSTeX', ^|\bull| for ^^{square bullet} `\bull', |\dn| and |\up| for
`\dn' and `\up', |\|\| and |\]| for `\|' and ^^{visible space} `\]'.  ^^{the
visible space symbol} To typeset
\begindisplay
$3\pt$ of \<stuff>, $\oct{105}=69$, $\hex{69}=105$, \cstok{wow}
\enddisplay
the manuscript says ^^{lxix}^^{octal constant}^^{hexadecimal constant}
^^{angle brackets} % cf. The Hacker's Dictionary
\begintt
$3\pt$ of \<stuff>, $\oct{105}=69$, $\hex{69}=105$, \cstok{wow}
\endtt
using the macros |\pt|, |\<|, |\oct|, |\hex|, and |\cstok|.
\beginlines
|\def\MF{{\manual META}\-{\manual FONT}}|
|\def\AmSTeX{$\cal A\kern-.1667em\lower.5ex\hbox{$\cal M$}\kern-.075em|
|  S$-\TeX}|
|\def\bull{\vrule height.9ex width.8ex depth-.1ex \relax} % square bullet|
|\def\SS{{\it SS}} % scriptscript style|
|\def\dn{\leavevmode\hbox{\tt\char'14}} % downward arrow|
|\def\up{|^|\leavevmode||\hbox{\tt\char'13}} % upward arrow|
|\def\|\||{\leavevmode\hbox{\tt\char`\|\||}} % vertical line|
|\def\]{\leavevmode\hbox{\tt\char`\ }} % visible space|
\smallbreak
|\def\pt{\,{\rm pt}} % units of points, in math formulas|
|\def\<#1>{\leavevmode\hbox{$\langle$#1\/$\rangle$}} % syntactic quantity|
|\def\oct#1{\hbox{\rm\'{}\kern-.2em\it#1\/\kern.05em}} % octal constant|
|\def\hex#1{\hbox{\rm\H{}\tt#1}} % hexadecimal constant|
|\def\cstok#1{\leavevmode\thinspace\hbox{|^|\vrule||\vtop{\vbox{\hrule\kern1pt|
|        \hbox{\vphantom{\tt/}\thinspace{\tt#1}\thinspace}}|
|      \kern1pt|^|\hrule||}\vrule}\thinspace} % control sequence token|
\endlines

^{Displays} in this book are usually indented rather than centered,
and they usually involve text rather than mathematics. The
|manmac| format makes such displays convenient by introducing
two macros called |\begindisplay| and |\enddisplay|; there's
also a pair of macros |\begintt| and |\endtt| for displays that
are entirely in ^{typewriter type}. The latter displays are copied
^{verbatim} from the manuscript file, without interpreting
symbols like |\| or |$| in any special way. For example,
part of the paragraph above was typed as follows:
\begintt
... To typeset
\begindisplay
$3\pt$ of \<stuff>, $\oct{105}=69$, $\hex{69}=105$, \cstok{wow}
\enddisplay
the manuscript says
\begintt
$3\pt$ of \<stuff>, $\oct{105}=69$, $\hex{69}=105$, \cstok{wow}
|char`|\endtt
using the macros ||\pt||, ||\<||, ||\oct||, ||\hex||, and ||\cstok||.
\endtt
(The last line of this example illustrates the fact
that verbatim typewriter text can be obtained within a
paragraph by using vertical lines as brackets.) \ The |\begindisplay|
macro is actually more general than you might expect from this example;
it allows multiline displays, with ^|\cr| following each line, and
it also allows local definitions (which apply only within the display) to
be specified immediately after |\begindisplay|.
\beginlines
|\outer\def\begindisplay{\obeylines\startdisplay}|
|{\obeylines\gdef\startdisplay#1|
|  {\catcode`\^^M=5$$#1\halign|^|\bgroup||\indent##\hfil&&\qquad##\hfil\cr}}|
|\outer\def\enddisplay{|^|\crcr|^|\egroup||$$}|^^{dollardollar}
\smallbreak
^|\chardef|^|\other||=12|
|\def\ttverbatim{\begingroup \catcode`\\=\other \catcode`\{=\other|
|  \catcode`\}=\other \catcode`\$=\other \catcode`\&=\other|
|  \catcode`\#=\other \catcode`\%=\other \catcode`\~=\other|
|  \catcode`\_=\other \catcode`\^=\other|
|  \obeyspaces \obeylines \tt}|
|{|^|\obeyspaces||\gdef {\ }} % \obeyspaces now gives \ , not \space|
\smallbreak
|\outer\def\begintt{$$\let\par=\endgraf \ttverbatim \parskip=0pt|
|  \catcode`\|\||=0 |^|\rightskip||=-5pc \ttfinish}|
|{\catcode`\|\||=0 |\||catcode`|\||\=\other % |\|%
  | is temporary escape character|
|  |\||obeylines % end of line is active|
|  |\||gdef|\||ttfinish#1^^M#2\endtt{#1|\||vbox{#2}|\||endgroup$$}}|
\smallbreak
^|\catcode||`\|\||=|^|\active|
|{\obeylines\gdef|\||{\ttverbatim\spaceskip=\ttglue\let^^M=\ \let|\|%
  |=\endgroup}}|\kern-.3pt
\endlines
These macros are more subtle than the others in this appendix, and they
deserve careful study because they illustrate how to disable \TeX's
normal formatting. The `\|' character is normally active (category~13)
in |manmac| format, and its appearance causes the |\ttverbatim| macro to
make all of the other unusual characters into normal symbols (category~12).
However, within the scope of\/ |\begintt...\endtt| a vertical line is an
^{escape character} (category~0); this permits an escape out of verbatim mode.

The |\begintt| macro assumes that a comparatively small amount of
text will be displayed; the verbatim lines are put into a vbox, so
that they cannot be broken between pages. A different approach has
been used for most of the typewriter copy in this appendix and in
Appendix~B\null: Material that is quoted from a format file is
delimited by |\beginlines| and |\endlines|, between which it
is possible to give commands like `^|\smallbreak|' to help with
spacing and page breaking. The |\beginlines| and |\endlines| macros
also insert rules, fore and aft:
\beginlines
|\def\beginlines{\par\begingroup\nobreak\medskip\parindent=0pt|
\nobreak
|  |^|\hrule||\kern1pt\nobreak \obeylines |^|\everypar||{\strut}}|
|\def\endlines{\kern1pt\hrule\endgroup\medbreak\noindent}|
\endlines
For example, the previous three lines were typeset by the specification
\begintt
\beginlines
||\def\beginlines{\par\begingroup\nobreak\medskip\parindent=0pt||
\nobreak
||  \hrule\kern1pt\nobreak \obeylines \everypar{\strut}}||
||\def\endlines{\kern1pt\hrule\endgroup\medbreak\noindent}||
\endlines
\endtt
A ^{strut} is placed in each line so that the rules will be positioned
properly. The |manmac| format also has macros |\beginmathdemo...\endmathdemo|
that were used to produce examples of mathematics in Chapters 16--19,
|\beginsyntax...\endsyntax| for the formal syntax in Chapters 24--26,
|\beginchart...\endchart| for the font tables in Appendices C and~F\null, etc.;
those macros are comparatively simple and they need not be shown here.

Exercises are specified by an |\exercise| macro; for example, the first
exercise in Chapter~1 was generated by the following lines in the
manuscript:
\begintt
\exercise After you have mastered the material in this book,
what will you be: a \TeX pert, or a \TeX nician?
\answer A \TeX nician (underpaid); sometimes also called
a \TeX acker.
\endtt
Notice that the |\answer| is given immediately after each exercise;
that makes it easy to insert new exercises or to delete old ones,
without keeping track of exercise numbers. Exercises that are dangerous
or doubly dangerous are introduced by the macros |\dangerexercise|
and~|\ddangerexercise|.
\beginlines
|\outer\def\exercise{\medbreak \global\advance\exno by 1|
|  \noindent|^|\llap||{\manual\char'170\rm\kern.15em}% triangle in margin|
|  {\ninebf EXERCISE \bf\chapno.\the\exno}\par\nobreak\noindent}|
|\def\dexercise{\global\advance\exno by 1|
|  \llap{\manual\char'170\rm\kern.15em}% triangle in indented space|
|  {\eightbf EXERCISE \bf\chapno.\the\exno}\hfil\break}|
|\outer\def\dangerexercise{\d@nger \dexercise}|
|\outer\def\ddangerexercise{\dd@nger \dexercise}|
\endlines
(The last two lines use |\d@nger| and |\dd@nger|, which are non-|\outer|
equivalents of\/ |\danger| and |\ddanger|; such duplication is necessary
because control sequences of type ^|\outer| cannot appear within a |\def|.)

The |\answer| macro copies an answer into a file called
|answers.tex|; then Appendix~A reads this
file by saying `^|\immediate|^|\closeout||\ans| |\ninepoint|
^|\input| |answers|'. Each individual answer ends with a blank line;
thus, |\par| must be used between the paragraphs of a long answer.
\beginlines
^|\newwrite||\ans|
|\immediate|^|\openout||\ans=answers % file for answers to exercises|
|\outer\def\answer{\par\medbreak|
|  \immediate|^|\write||\ans{}|
|  \immediate\write\ans{\string\ansno\chapno.|^|\the||\exno:}|
|  \copytoblankline}|
|\def\copytoblankline{\begingroup\setupcopy\copyans}|
|\def\setupcopy{\def\do##1{\catcode`##1=\other}\dospecials|
|  \catcode`\|\||=\other \obeylines}|
|{\obeylines \gdef\copyans#1|
|  {\def\next{#1}%|
|  \ifx\next\empty\let\next=\endgroup %|
|  \else\immediate\write\ans{\next} \let\next=\copyans\fi\next}}|
\endlines
Notice the use of\/ ^|\dospecials| here, to set up the ^{verbatim copying}.
The |\ttverbatim| macro could have invoked |\dospecials| in the same way;
^^{efficiency}
but |\ttverbatim| is used quite frequently, so it was streamlined for speed.

The remaining macros in |manmac| format are designed to help in producing
a good ^{index}. When a paragraph contains a word or group of words that
deserve to be indexed, the manuscript indicates this by inserting
|^{...}|; for example, the first sentence of the present paragraph
actually ends with `|a good ^{index}|'. This caused an appropriate
entry to be written onto a file |index.tex| when \TeX\ was typesetting
the page; it also put the word `{\sevenrm index}' into the margin
of the proofsheets, so that the author could remember what
had been marked for indexing without looking into the manuscript file.
Indexing with the |^{...}| notation doesn't change \TeX's behavior in
any essential way; thus, the word `index' appears
in the text as well as in the index. ^^{strut}
\beginlines
^|\newwrite||\inx|
^|\immediate|^|\openout||\inx=index % file for index reminders|
|\def\marginstyle{\sevenrm % marginal notes are in 7-point type|
|  \vrule height6pt depth2pt width0pt \relax} % a strut for \insert\margin|
\endlines

Sometimes it is desirable to index words that don't actually appear on the
page; the notation |^^{...}| stands for a ``silent'' index entry, and
spaces are ignored after the closing `|}|' in such a case. For example,
Appendix~I lists page~1 under `beauty', even though page~1 contains only
the word `beautiful'; the manuscript achieves this by saying
`|beautiful ^^{beauty}|'.  \ (The author felt that it was important to
index `beauty' because he had already indexed `truth'.)

It's not difficult to make |^| into an active character that produces
such index entries, while still retaining its use for superscripts in
math formulas, because ^|\ifmmode| can be used to test whether a
control sequence is being used in math mode. However, |manmac|'s use
of\/ |^| as an active character means that |^^M| cannot be used to refer
^^{hat as an active character} ^^{hat hat}
to a \<return> character. Fortunately the |^^M| notation isn't needed
except when the formatting macros themselves are being defined.

The following macros set things up so that |^| and |^^| are respectively
converted to |\silentfalse\xref| and |\silenttrue\xref|, outside of
math mode:
\beginlines
^|\newif||\ifsilent|
|\def\specialhat{\ifmmode\def\next{^}\else\let\next=\beginxref\fi\next}|
|\def\beginxref{|^|\futurelet||\next\beginxrefswitch}|
|\def\beginxrefswitch{\ifx\next\specialhat\let\next=\silentxref|
|  \else\silentfalse\let\next=\xref\fi \next}|
|\catcode`\^=\active \let ^=\specialhat|
|\def\silentxref^{\silenttrue\xref}|
\endlines

Entries in the index aren't always words in roman type; they might require
special typesetting conventions. For example, there are
hundreds of items in Appendix~I that are preceded by a backslash and
set in typewriter type. The |manmac| format makes it easy to produce
such entries by typing, e.g., `|^|\||\immediate|\|' instead of
`|^{|\||\immediate|\||}|'. In this case the backslash is not written
onto the |index| file, because it would interfere with alphabetization
of the entries; a code number is written out so that the backslash can
be reinstated after the index has been sorted. The code number also
is used to put the entry in typewriter type.

The indexing macros of |manmac| produce entries of four kinds, which are
assigned to codes 0,~1, 2, and~3. Code~0 applies when the argument is enclosed
in braces, e.g., `|^{word}|'\thinspace; code~1 applies when the argument
is enclosed in vertical lines and there's no backslash, e.g.,
`|^|\||plus|\|'\thinspace; code~2 is similar but with a backslash, e.g.,
`|^|\||\par|\|'\thinspace; code~3 applies when the argument is enclosed in
angle brackets, e.g., `|^\<stuff>|'.  The four example entries in the previous
sentence will be written on file |index.tex| in the form\strut
\begintt
word !0 123.
plus !1 123.
par !2 123.
stuff !3 123.
|kern-3pt
\endtt
if they appear on page 123 of the book.
\beginlines
|\chardef\bslash=`\\ % \bslash makes a backslash (in tt fonts)|
|\def\xref{\futurelet\next\xrefswitch} % branch on the next character|
|\def\xrefswitch{\begingroup\ifx\next|\||\aftergroup\vxref|
|  \else\ifx\next\<\aftergroup\anglexref|
|    \else\aftergroup\normalxref \fi\fi \endgroup}|
|\def\vxref|\||{\catcode`\\=\active \futurelet\next\vxrefswitch}|
|\def\vxrefswitch#1|\||{\catcode`\\=0|
|  \ifx\next\empty\def\xreftype{2}%|
|    \def\next{{\tt\bslash\text}}% code 2, |\||\arg|\|
|  \else\def\xreftype{1}\def\next{{\tt\text}}\fi % code 1, |\||arg|\|
|  \edef\text{#1}\makexref}|
|{|^|\catcode||`\|\||=0 \catcode`\\=\active |\||gdef\{}}|
|\def\anglexref\<#1>{\def\xreftype{3}\def\text{#1}%|
|  \def\next{\<\text>}\makexref} % code 3, \<arg>|
|\def\normalxref#1{\def\xreftype{0}\def\text{#1}\let\next=\text\makexref}|
\endlines

Indexing is suppressed unless the |proofmode| switch is set to true,
since material is gathered for the index only during trial runs---not
on the triumphant occasion when the book is finally being printed.
\beginlines
|\newif\ifproofmode|
|\proofmodetrue % this should be false when making camera-ready copy|
|\def\makexref{\ifproofmode\insert\margin{\hbox{\marginstyle\text}}%|
|   |^|\xdef||\writeit{|^|\write||\inx{\text\space!\xreftype\space|
|     |^|\noexpand|^|\number||\pageno.}}\writeit|
|   \else\ifhmode\kern0pt \fi\fi|
|  \ifsilent|^|\ignorespaces||\else\next\fi}|
\endlines
(The ^|\insert| suppresses ^{hyphenation} when proofs are being checked;
a ^|\kern||0pt| is therefore emitted to provide consistent behavior
in the other case.)

The material that accumulates on file |index.tex| gives a good first
approximation to an index, but it doesn't contain enough information to do
the whole job; a topic often occurs on several pages, but only the first
of those pages is typically listed in the file.  The author ^^{Knuth}
prefers not to generate indexes automatically; he likes to reread his
books as he checks the cross-references, thereby having the opportunity to
rethink everything and to catch miscellaneous errors before it is too
late. As a result, his books tend to be delayed, but the indexes tend to
be pretty good.  Therefore he designed the indexing scheme of |manmac| to
provide only the clues needed to make a real index. On the other hand,
it would be possible to extend the macros above and to obtain a comprehensive
system that generates an excellent index with no subsequent human intervention;
see, for example, ``An indexing facility for \TeX'' by Terry ^{Winograd}
and Bill ^{Paxton}, in {\sl ^{TUGboat}\/ \bf1} (1980), A1--A12.

The |manmac| macros have now been fully presented; we shall close this
appendix by presenting one more example of their use. Chapter~27 mentions
the desirability of creating a long book in small parts, by using a
``^{galley}'' file. The author adopted that strategy for {\sl The \TeX book},
entering each chapter into a small file |galley.tex| that looked like this:
\beginlines
|\input manmac|
|\tenpoint|
|\pageno=800|
|\def\rhead{Experimental Pages for The \TeX book}|
|\def\chapno{ X}|
|{\catcode`\%=12 \immediate\write\ans{% Answers for galley proofs:}}|
\qquad\vdots
\<new text being tested, usually an entire chapter>
\qquad\vdots
| |
|% that blank line will stop an unfinished \answer|
|\immediate\closeout\ans|
|\vfill\eject|
|\ninepoint \input answers % typeset the new answers, if any|
|\bye|
\endlines

\endchapter

It is much easier to use macros than to define them.
$\ldots$
The use of macro libraries, in fact, mirrors almost exactly
the use of subroutine libraries for programming languages.
There are the same levels of specialization,
from publicly shared subroutines
to special subroutines within a single program,
and there is the same need for a programmer
with particular skills to define the subroutines.
\author PETER ^{BROWN}, {\sl Macro Processors\/} (1974) % p10

\bigskip

The ^{epigraph} is among the most delightful of scholarly habits.
Donald ^{Knuth}'s work on fundamental algorithms would be
just as important if he hadn't begun with a quotation
from Betty ^{Crocker}, but not so enjoyable.
Part of the fun of an epigraph is turning a source to an unexpected use.
\author MARY-CLAIRE ^{VAN LEUNEN}, {\sl A Handbook for Scholars\/} (1978)
 % page 53. [But it was McCall's, not Betty Crocker]

\eject
\beginchapter Appendix F. Font Tables

The purpose of this appendix is to summarize the chief characteristics of
the ^{Computer Modern} typefaces.  \TeX\ is able to typeset
documents with any fonts, having any arrangement of characters; the
fonts and layouts to be described here are the particular ones that
correspond to plain~\TeX\ format, i.e., to the macros in Appendix~B\null.
\ (Complete information about the Computer Modern family, including the
\MF\ programs that draw the characters, can be found in the author's book
{\sl Computer Modern Typefaces}.)
^^{METAFONT}

The first pages of this appendix show what the fonts contain; the last
pages show what the symbols are called when they're used in math formulas.
\ (See Appendix~B for the conventions that apply in non-mathematical text.)

There are exactly 128 different characters in each of the Computer
Modern fonts, although \TeX\ can work up to~256 characters per font.
The text fonts are laid out as shown in the table below, which illustrates
font |cmr10| (Computer Modern Roman 10~point). Thus, for example,
if you ask for ^|\char||'35| when ^|cmr10| is the current font, you
get the symbol \char'35. These text fonts include the ^{ligatures} and
^{accents} described in Chapter~9; each symbol that happens to be
a visible ASCII character appears in its ASCII position.
 Some of the ASCII symbols (namely
|"|~|<|~|>|~|\|~|_|~|{|~\|~|}|) are not included because
they don't occur in normal printer's fonts. If you mistakenly
type |"|, you get "; and |<| outside of math mode yields <\thinspace!
\ Incidentally, the ten ^{digits} all have width |0.5em|.

\medskip\vfill
\noindent {\bf Figure 1.\enspace}Text font layout,
showing |cmr10| (^|\rm|, |\textfont0|).
\beginchart\tenrm
\normalchart
\endchart

\noindent Plain \TeX\ makes use of sixteen basic fonts:
$$\halign{\indent$\nulldelimiterspace=0pt
  \left.\ninepoint\vcenter{#}\,\hfil\right\}$ &#\hfil\cr
\halign{\strut\tt cm# \ \hfil&(Computer Modern #)\hfil\cr
r10&Roman 10 point\cr
r7&Roman 7 point\cr
r5&Roman 5 point\cr
bx10&Bold Extended 10 point\cr
bx7&Bold Extended 7 point\cr
bx5&Bold Extended 5 point\cr
sl10&Slanted Roman 10 point\cr
ti10&Text Italic 10 point\cr}&text\cr
\halign{\strut\tt cm# \ \hfil&(Computer Modern #)\hfil\cr
tt10&Typewriter Type 10 point\cr
mi10&Math Italic 10 point\cr
mi7&Math Italic 7 point\cr
mi5&Math Italic 5 point\cr
sy10&Math Symbols 10 point\cr
sy7&Math Symbols 7 point\cr
sy5&Math Symbols 5 point\cr
ex10&Math Extension 10 point\cr}&special\cr}$$
The first eight of these all have essentially the same layout;
but |cmr5| needs no ligatures, and many of the symbols of |cmti10|
have different shapes.
For example, the ^{ampersand} becomes an `^{E.T.}', and the
^{dollar} changes to ^{pound} ^{sterling}:

\medskip\vfill
\noindent {\bf Figure 2.\enspace}Text font layout,
showing |cmti10| (^|\it|).
\beginchart\tenit
\normalchart
\endchart

\noindent The ^{typewriter font} ^|cmtt10| is almost like the fonts for
ordinary text, but it includes all of the visible ASCII characters, in
their correct positions. It also has vertical arrows \up\ and
\dn, as well as an undirected single quote mark, {\tt\char'15}.
Fourteen of the~128 positions are changed from the normal text layout
conventions,
namely codes \oct{013}--\oct{017}, \oct{040}, \oct{042}, \oct{074},
\oct{076}, \oct{134}, \oct{137}, and \oct{173}--\oct{175}. All of the
ligatures are absent, except for the Spanish |!`| and |?`|. \ (The
characters for Spanish
ligatures appear in different positions, but that makes no difference
to the user, because each font tells \TeX\ where to locate its own
ligatures.) \ The Polish \l, the d\.ot accent, and the
long Hung\H arian umlaut have disappeared to make room for new symbols.
In a sense, positions \oct{052} and \oct{055} also differ from the normal
text conventions: The asterisk is not up as high as usual, and the hyphen is
just like a minus sign.

Each character in |cmtt10| has the same width, namely |0.5em|;
the spaces between words also have this width, and they will not
stretch or shrink. \TeX\ puts two spaces at the end of each sentence
when you are typesetting with a typewriter font. \ (These spacing
conventions can be changed by assigning nonzero values to
^|\spaceskip| and ^|\xspaceskip|; or you can assign new values to the
|\fontdimen| parameters, which will be described shortly.)

\medskip\vfill
\noindent {\bf Figure 3.\enspace}Typewriter text font layout,
showing |cmtt10| (^|\tt|).
\beginchart\tentt
\normalchart
\endchart

\noindent You can see at a glance that the math italic font, ^|cmmi10|,
is quite different from ^{text italic}. It contains lowercase Greek letters
as well as uppercase ones; this, of course, is mathematicians' ^{Greek},
not a text font that would be suitable for typesetting classical
Greek literature. And if you look closely at the non-Greek italic
letters, you will find that their proportions and spacing have been
changed from |cmti10| to make them work better in \TeX's mathematics mode.

Some special unslanted characters appear in positions \oct{050}--\oct{077}
and \oct{133}--\oct{137}, including ``^{oldstyle numerals}'':
`|$\mit1984$|' and `|$\oldstyle1984$|' both yield `$\mit1984$'. Some of the
characters are intended to be combined with others; for example, \oct{054}
forms the first part of the symbol `$\hookrightarrow$'. \ (See the definition
of\/ ^|\hookrightarrow| in Appendix~B\null.) \ This portion of the font doesn't
deserve the name math italic; it's really a resting place for characters
that don't fit anywhere else. \ (The author didn't want to leave any places
unfilled, since that would tempt people to create incompatible
ways to fill them.)

Plain \TeX\ takes its ^{comma}, ^{period}, and ^{slash} from |cmmi10| in
math mode, so that appropriate kerning will be computed in certain
formulas that would otherwise be spaced poorly. For the correct
positioning of math accents with this font, you should set its
^|\skewchar| to \oct{177}.

\medskip\vfill
\noindent {\bf Figure 4.\enspace}Math italic font layout,
showing |cmmi10| (^|\mit|, |\textfont1|).
\beginchart\teni
\normalchart
\endchart

\noindent When \TeX\ typesets mathematics it assumes that family~0 contains
normal roman fonts and that families 1, 2, and~3 contain math italic,
math symbol, and math extension fonts. The special characters in these
fonts are usually given symbolic names by a |\mathchardef|
instruction, which assigns a hexadecimal code to the symbol.
This code has four digits, where the first tells what kind of symbol
is involved, the second specifies the family, and the other two give the font
position. For example,
\begintt
\mathchardef\ll="321C
\endtt
says that ^|\ll| is character \hex{1C} of the math symbol font (family~2),
and that it's a ``relation'' (class~3). A complete list of the symbolic names
provided by the plain \TeX\ format appears later in this appendix.

\smallskip
Font ^|cmsy10| is plain \TeX's math symbol font, and it contains 128
symbols laid out as shown below. Its ^|\skewchar| should be set to
\oct{060} so that math accents will be positioned properly over the
^{calligraphic capital letters}.

\medskip\vfill
\noindent {\bf Figure 5.\enspace}Math symbol font layout,
showing |cmsy10| (^|\cal|, |\textfont2|).
\beginchart\tensy
\normalchart
\endchart

The final font of plain \TeX\ is ^|cmex10|, which includes large symbols
and pieces that can be used to build even larger ones. For example,
arbitrarily large left parentheses can be constructed by putting
\oct{060} at the top and \oct{100} at the bottom,
and by using as many copies of \oct{102} as necessary in the middle.
Large ^{square root signs} are made from \oct{164}, \oct{165}, and
\oct{166}; large left braces have four component parts:
\oct{070}, \oct{072}, \oct{074}, \oct{076}.

\medskip\vfill
\noindent {\bf Figure 6.\enspace}^{Math extension font} layout,
showing |cmex10| (|\textfont3|).
\beginchart\tenex
\normalchart
\endchart

\noindent When \TeX\ ``loads'' a font into its memory, it doesn't look at
the actual shapes of the characters; it only loads the {\sl ^{font metric}\/}
information (e.g., |cmr10.tfm|), which includes the heights, widths,
depths, and italic corrections, together with information about ligatures
and kerning.  Furthermore, the metric information that comes with a font
like |cmex10| tells \TeX\ that certain characters form a series; for
example, all of the left parentheses are linked together in order of
increasing size: \oct{000}, \oct{020}, \oct{022}, and \oct{040}, followed
by the extensible left parenthesis, which is $\oct{060}+[\oct{102}]^n+
\oct{100}$. Similarly, the two ^{summation signs} (\oct{120}, \oct{130})
and the three ^|\widehat| accents (\oct{142}, \oct{143}, \oct{144}) are
linked together.  Appendix~G explains how \TeX\ goes about choosing
particular sizes for math delimiters, math operators, and math accents.

\smallskip
Each font also has at least seven ^|\fontdimen| parameters, which have the
following significance and typical values (rounded to two decimal places):
$$\halign to\hsize{#\tabskip0pt plus 10pt&#\hfil&
$\hfil#\pt$&
$\hfil#\pt$&
$\hfil#\pt$&
$\hfil#\pt$&
$\hfil#\pt$&
$\hfil#\pt$\tabskip0pt\cr
\hidewidth\#\hidewidth&Meaning&
\omit\hidewidth Value in |cmr10| &
\omit\hfil|cmbx10| &
\omit\hfil|cmsl10| &
\omit\hfil|cmti10| &
\omit\hfil|cmtt10| &
\omit\hfil|cmmi10| \cr
\noalign{\vskip2pt}
1&slant per pt&0.00&0.00&0.17&0.25&0.00&0.25\cr
2&^{interword space}&3.33&3.83&3.33&3.58&5.25&0.00\cr
3&interword stretch&1.67&1.92&1.67&1.53&0.00&0.00\cr
4&interword shrink&1.11&1.28&1.11&1.02&0.00&0.00\cr
5&^{x-height}&4.31&4.44&4.31&4.31&4.31&4.31\cr
6&^{quad} width&10.00&11.50&10.00&10.22&10.50&10.00\cr
7&extra space&1.11&1.28&1.11&1.02&5.25&0.00\cr}$$
The ^{slant} parameter is used to position accents;
the next three parameters define interword spaces when text is being typeset;
the next two define the font-oriented dimensions |1ex| and |1em|;
^^|ex| ^^|em|
and the last is the additional amount that is added to interword spaces
at the end of sentences
(i.e., when ^|\spacefactor| is 2000 or~more and ^|\xspaceskip| is zero).
When a font is ^{magnified} (using `^|at|' or `^|scaled|'), all of the
parameters except the slant are subject to magnification at the time the
font is loaded into \TeX's memory.

Notice that |cmmi10| has zero spacing. This is the mark of a font
that is intended only for mathematical typesetting; ^^{math fonts}
the rules in Appendix~G state that the italic correction is added
between adjacent characters from such fonts.

Math symbol fonts (i.e., fonts in family 2) are required to have at least
22~^|\fontdimen| parameters instead of the usual seven; similarly, math
extension fonts must have at least~13. The significance of these
additional parameters is explained in Appendix~G\null. If you want to increase
the number of parameters past the number that actually appear in a font's
metric information file, you can assign new values immediately after that font
has been loaded.  For example, if some font |\ff| with seven parameters
has just entered \TeX's memory, the command |\fontdimen13\ff=5pt| will set
parameter number~13 to $5\pt$; the intervening parameters, numbers 8--12,
will be set to zero. You can even give more than seven parameters to
^|\nullfont|, provided that you assign the values before any actual fonts
have been loaded.

\goodbreak
Now that the font layouts have all been displayed, it's time to
consider the names of the various mathematical symbols.
^^{symbols in math, table}
Plain \TeX\ defines more than 200 control sequences by which you can refer
to math symbols without having to find their numerical positions in the
layouts. It's generally best to call a symbol by its name, for then you can
easily adapt your manuscripts to other fonts, and your manuscript will
be much more readable.

The symbols divide naturally into groups based on their mathematical
class (Ord, Op, Bin, Rel, Open, Close, or Punct), so we shall follow
that order as we discuss them. N.B.: Unless otherwise stated, math symbols
are available only in math modes. For example, if you say `|\alpha|' in
horizontal mode, \TeX\ will report an error and try to insert a |$| sign.

\def\beginsymbols{$$\displayindent=16pt
  \halign\bgroup
  &\qquad\hbox to10pt{\hss$##$\hss}\enspace&\hbox to80pt{##\hss}\cr}
\let\endsymbols=\enddisplay

\subsection Lowercase Greek letters.
^^|\alpha|^^|\iota|^^|\varrho|
^^|\beta|^^|\kappa|^^|\sigma|
^^|\gamma|^^|\lambda|^^|\varsigma|
^^|\delta|^^|\mu|^^|\tau|
^^|\epsilon|^^|\nu|^^|\upsilon|
^^|\varepsilon|^^|\xi|^^|\phi|
^^|\zeta|^^|\varphi|
^^|\eta|^^|\pi|^^|\chi|
^^|\theta|^^|\varpi|^^|\psi|
^^|\vartheta|^^|\rho|^^|\omega|
\beginsymbols
\alpha&|\alpha|&\iota&|\iota|&\varrho&|\varrho|\cr
\beta&|\beta|&\kappa&|\kappa|&\sigma&|\sigma|\cr
\gamma&|\gamma|&\lambda&|\lambda|&\varsigma&|\varsigma|\cr
\delta&|\delta|&\mu&|\mu|&\tau&|\tau|\cr
\epsilon&|\epsilon|&\nu&|\nu|&\upsilon&|\upsilon|\cr
\varepsilon&|\varepsilon|&\xi&|\xi|&\phi&|\phi|\cr
\zeta&|\zeta|&o&|o|&\varphi&|\varphi|\cr
\eta&|\eta|&\pi&|\pi|&\chi&|\chi|\cr
\theta&|\theta|&\varpi&|\varpi|&\psi&|\psi|\cr
\vartheta&|\vartheta|&\rho&|\rho|&\omega&|\omega|\cr
\endsymbols
There's no ^|\omicron|, because it would look the same as |o|. Notice
that the letter |\upsilon|~($\upsilon$) is a bit wider than |v|~($v$); both
of them should be distinguished from |\nu|~($\nu$). Similarly,
|\varsigma|~($\varsigma$) should not be confused with |\zeta|~($\zeta$).
It turns out that |\varsigma| and |\upsilon| are almost never used in
math formulas; they are included in plain \TeX\ primarily because they are
sometimes needed in short Greek citations (cf.~Appendix~J).

\subsection Uppercase Greek letters.
^^|\Gamma|^^|\Xi|^^|\Phi|
^^|\Delta|^^|\Pi|^^|\Psi|
^^|\Theta|^^|\Sigma|^^|\Omega|
^^|\Lambda|^^|\Upsilon|
\beginsymbols
\Gamma&|\Gamma|&\Xi&|\Xi|&\Phi&|\Phi|\cr
\Delta&|\Delta|&\Pi&|\Pi|&\Psi&|\Psi|\cr
\Theta&|\Theta|&\Sigma&|\Sigma|&\Omega&|\Omega|\cr
\Lambda&|\Lambda|&\Upsilon&|\Upsilon|\cr
\endsymbols
The other Greek capitals appear in the roman alphabet
(^|\Alpha|${}\equiv{}$|{\rm A}|, ^|\Beta|${}\equiv{}$|{\rm B}|, etc.).
It's conventional to use unslanted letters for uppercase Greek, and
slanted letters for lowercase Greek; but you can obtain
$(\mit\Gamma,\Delta,\ldots,\Omega)$ by typing
|$({\mit\Gamma},| |{\mit\Delta},| |\ldots,| |{\mit\Omega})$|. ^^|\mit|

\subsection Calligraphic capitals. To get the letters ^^{calligraphic letters}
$\cal A\ldots Z$ that appear in Figure~5, type |${\cal A}\ldots{\cal Z}$|.
Several other alphabets are also used with mathematics (notably ^{Fraktur},
^{script}, and ``^{blackboard bold}''); they don't come with plain \TeX,
but more elaborate formats like \AmSTeX\ do provide them.

\subsection Miscellaneous symbols of type Ord.
^^|\aleph|^^|\prime|^^|\forall|
^^|\hbar|^^|\emptyset|^^|\exists|
^^|\imath|^^|\nabla|^^|\neg|
^^|\jmath|^^|\surd|^^|\flat|
^^|\ell|^^|\top|^^|\natural|
^^|\wp|^^|\bot|^^|\sharp|
^^|\Re|^^{escvert}^^|\clubsuit|
^^|\Im|^^|\angle|^^|\diamondsuit|
^^|\partial|^^|\triangle|^^|\heartsuit|
^^|\infty|^^|\backslash|^^|\spadesuit|
^^{Weierstrass, see wp}
\beginsymbols
\aleph&|\aleph|&\prime&|\prime|&\forall&|\forall|\cr
\hbar&|\hbar|&\emptyset&|\emptyset|&\exists&|\exists|\cr
\imath&|\imath|&\nabla&|\nabla|&\neg&|\neg|\cr
\jmath&|\jmath|&\surd&|\surd|&\flat&|\flat|\cr
\ell&|\ell|&\top&|\top|&\natural&|\natural|\cr
\wp&|\wp|&\bot&|\bot|&\sharp&|\sharp|\cr
\Re&|\Re|&\Vert&|\|\|&\clubsuit&|\clubsuit|\cr
\Im&|\Im|&\angle&|\angle|&\diamondsuit&|\diamondsuit|\cr
\partial&|\partial|&\triangle&|\triangle|&\heartsuit&|\heartsuit|\cr
\infty&|\infty|&\backslash&|\backslash|&\spadesuit&|\spadesuit|\cr
\endsymbols
The ^{dotless letters} |\imath| and |\jmath| should be used when $i$ and~$j$
are ^{accent}ed; for example, |$\hat\imath$| yields $\hat\imath$.
The |\prime| symbol is intended for use in subscripts and superscripts,
as explained in Chapter~16, so you usually see it in a smaller size.
On the other hand, the |\angle| symbol has been built up from other pieces;
it does not get smaller when it appears in a subscript or superscript.

\subsection Digits. To get italic ^{digits} {\it0123456789}, say
|{\it0123456789}|; to get boldface digits {\bf0123456789}, say
|{\bf0123456789}|; to get oldstyle digits {\oldstyle0123456789}, say
|{\oldstyle0123456789}|. These conventions work also outside of math mode.

\subsection ``Large'' operators. The following symbols come in two sizes,
for text and display styles:
^^|\sum|^^|\bigcap|^^|\bigodot|
^^|\prod|^^|\bigcup|^^|\bigotimes|
^^|\coprod|^^|\bigsqcup|^^|\bigoplus|
^^|\int|^^|\bigvee|^^|\biguplus|
^^|\oint|^^|\bigwedge|
$$\displayindent=16pt \openup3pt
\halign{&\qquad\hbox to10pt{\hss$#$\hss}\enspace&
  \hbox to10pt{\hss$\displaystyle#$\hss}\enspace&
  \hbox to60pt{#\hss}\cr
\sum&\sum&|\sum|&\bigcap&\bigcap&|\bigcap|&
  \bigodot&\bigodot&|\bigodot|\cr
\prod&\prod&|\prod|&\bigcup&\bigcup&|\bigcup|&
  \bigotimes&\bigotimes&|\bigotimes|\cr
\coprod&\coprod&|\coprod|&\bigsqcup&\bigsqcup&|\bigsqcup|&
  \bigoplus&\bigoplus&|\bigoplus|\cr
\int&\int&|\int|&\bigvee&\bigvee&|\bigvee|&
  \biguplus&\biguplus&|\biguplus|\cr
\oint&\oint&|\oint|&\bigwedge&\bigwedge&|\bigwedge|\cr
}$$
It is important to distinguish these large Op symbols from the similar but
smaller Bin symbols whose names are the same except for a `|big|' prefix.
Large operators usually occur at the beginning of a formula or subformula,
and they usually are subscripted; ^{binary operations} usually occur between
two symbols or subformulas, and they rarely are subscripted. For example,
\begindisplay
|$\bigcup_{n=1}^m(x_n\cup y_n)$|\qquad yields \qquad
$\bigcup_{n=1}^m(x_n\cup y_n)$
\enddisplay
The large operators |\sum|, |\prod|, |\coprod|, and |\int| should also be
distinguished from smaller symbols called |\Sigma|~($\Sigma$),
|\Pi|~($\Pi$), |\amalg|~($\amalg$), and |\smallint|~($\smallint$),
respectively; the ^|\smallint| operator is rarely used.

\subsection Binary operations. Besides $+$ and $-$, you can type
^^|\pm|^^|\cap|^^|\vee|
^^|\mp|^^|\cup|^^|\wedge|
^^|\setminus|^^|\uplus|^^|\oplus|
^^|\cdot|^^|\sqcap|^^|\ominus|
^^|\times|^^|\sqcup|^^|\otimes|
^^|\ast|^^|\triangleleft|^^|\oslash|
^^|\star|^^|\triangleright|^^|\odot|
^^|\diamond|^^|\wr|^^|\dagger|
^^|\circ|^^|\bigcirc|^^|\ddagger|
^^|\bullet|^^|\bigtriangleup|^^|\amalg|
^^|\div|^^|\bigtriangledown|
\beginsymbols
\pm&|\pm|&\cap&|\cap|&\vee&|\vee|\cr
\mp&|\mp|&\cup&|\cup|&\wedge&|\wedge|\cr
\setminus&|\setminus|&\uplus&|\uplus|&\oplus&|\oplus|\cr
\cdot&|\cdot|&\sqcap&|\sqcap|&\ominus&|\ominus|\cr
\times&|\times|&\sqcup&|\sqcup|&\otimes&|\otimes|\cr
\ast&|\ast|&\triangleleft&|\triangleleft|&\oslash&|\oslash|\cr
\star&|\star|&\triangleright&|\triangleright|&\odot&|\odot|\cr
\diamond&|\diamond|&\wr&|\wr|&\dagger&|\dagger|\cr
\circ&|\circ|&\bigcirc&|\bigcirc|&\ddagger&|\ddagger|\cr
\bullet&|\bullet|&\bigtriangleup&|\bigtriangleup|&\amalg&|\amalg|\cr
\div&|\div|&\bigtriangledown&|\bigtriangledown|\cr
\endsymbols
It's customary to say |$G\backslash| |H$| to denote double cosets
of $G$ by~$H$~($G\backslash H$), and |$p\backslash n$| to mean that
$p$ divides~$n$~($p\backslash n$); but |$X\setminus Y$| denotes
the elements of set~$X$ minus those of set~$Y$~($X\setminus Y$).
Both operations use the same symbol, but |\backslash| is
type Ord, while |\setminus| is type Bin (so \TeX\ puts
more space around it).

\subsection Relations. Besides $<$, $>$, and $=$, you can type
^^|\leq|^^|\geq|^^|\equiv|
^^|\prec|^^|\succ|^^|\approx|
^^|\preceq|^^|\succeq|^^|\propto|
^^|\ll|^^|\gg|^^|\asymp|
^^|\subset|^^|\supset|^^|\sim|
^^|\subseteq|^^|\supseteq|^^|\simeq|
^^|\sqsubseteq|^^|\sqsupseteq|^^|\cong|
^^|\in|^^|\ni|^^|\bowtie|
^^|\vdash|^^|\dashv|^^|\models|
^^|\smile|^^|\mid|^^|\doteq|
^^|\frown|^^|\parallel|^^|\perp|
\beginsymbols
\leq&|\leq|&\geq&|\geq|&\equiv&|\equiv|\cr
\prec&|\prec|&\succ&|\succ|&\sim&|\sim|\cr
\preceq&|\preceq|&\succeq&|\succeq|&\simeq&|\simeq|\cr
\ll&|\ll|&\gg&|\gg|&\asymp&|\asymp|\cr
\subset&|\subset|&\supset&|\supset|&\approx&|\approx|\cr
\subseteq&|\subseteq|&\supseteq&|\supseteq|&\cong&|\cong|\cr
\sqsubseteq&|\sqsubseteq|&\sqsupseteq&|\sqsupseteq|&\bowtie&|\bowtie|\cr
\in&|\in|&\ni&|\ni|&\propto&|\propto|\cr
\vdash&|\vdash|&\dashv&|\dashv|&\models&|\models|\cr
\smile&|\smile|&\mid&|\mid|&\doteq&|\doteq|\cr
\frown&|\frown|&\parallel&|\parallel|&\perp&|\perp|\cr
\endsymbols
The symbols |\mid| and |\parallel| define relations that use the same
characters as you get from \| and |\|\|; \TeX\ puts space around
them when they are relations.

\subsection Negated relations. Many of the relations just listed can be
negated or ``crossed out'' by prefixing them with ^|\not|, as follows:
\beginsymbols
\not<&|\not<|&\not>&|\not>|&\not=&|\not=|\cr
\not\leq&|\not\leq|&\not\geq&|\not\geq|&
  \not\equiv&|\not\equiv|\cr
\not\prec&|\not\prec|&\not\succ&|\not\succ|&
  \not\sim&|\not\sim|\cr
\not\preceq&|\not\preceq|&\not\succeq&|\not\succeq|&
  \not\simeq&|\not\simeq|\cr
\not\subset&|\not\subset|&\not\supset&|\not\supset|&
  \not\approx&|\not\approx|\cr
\not\subseteq&|\not\subseteq|&\not\supseteq&|\not\supseteq|&
  \not\cong&|\not\cong|\cr
\not\sqsubseteq&|\not\sqsubseteq|&\not\sqsupseteq&|\not\sqsupseteq|&
  \not\asymp&|\not\asymp|\cr
\endsymbols
The symbol |\not| is a relation character of width zero, so it will
overlap a relation that comes immediately after it. The positioning
isn't always ideal, because some relation symbols are wider than others;
for example, |\not\in| gives `$\not\in$', but it is preferable to
have a steeper cancellation, `$\notin$'. The latter symbol is
available as a special control sequence called ^|\notin|. The definition
of\/ |\notin| in Appendix~B indicates how similar symbols can be constructed.

\subsection Arrows. There's also another big class of relations, namely
^^{arrows}
^^|\leftarrow| ^^|\longleftarrow| ^^|\uparrow| ^^|\Leftarrow|
^^|\Longleftarrow| ^^|\Uparrow| ^^|\rightarrow| ^^|\longrightarrow|
^^|\downarrow| ^^|\Rightarrow| ^^|\Longrightarrow| ^^|\Downarrow|
^^|\leftrightarrow| ^^|\longleftrightarrow| ^^|\updownarrow|
^^|\Leftrightarrow| ^^|\Longleftrightarrow| ^^|\Updownarrow| ^^|\mapsto|
^^|\longmapsto| ^^|\nearrow| ^^|\hookleftarrow| ^^|\hookrightarrow|
^^|\searrow| ^^|\leftharpoonup| ^^|\rightharpoonup| ^^|\swarrow|
^^|\leftharpoondown| ^^|\rightharpoondown| ^^|\nwarrow|
^^|\rightleftharpoons|
those that point:
$$\displayindent=16pt
  \halign\bgroup
  \qquad\hbox to10pt{\hss$#$\hss}\enspace&\hbox to75pt{#\hss}&
  \qquad\hbox to10pt{\hss$#$\hss}\enspace&\hbox to85pt{#\hss}&
  \qquad\hbox to10pt{\hss$#$\hss}\enspace&\hbox to80pt{#\hss}\cr
\leftarrow&|\leftarrow|&\longleftarrow&|\longleftarrow|&
  \uparrow&|\uparrow|\cr
\Leftarrow&|\Leftarrow|&\Longleftarrow&|\Longleftarrow|&
  \Uparrow&|\Uparrow|\cr
\rightarrow&|\rightarrow|&\longrightarrow&|\longrightarrow|&
  \downarrow&|\downarrow|\cr
\Rightarrow&|\Rightarrow|&\Longrightarrow&|\Longrightarrow|&
  \Downarrow&|\Downarrow|\cr
\leftrightarrow&|\leftrightarrow|&\longleftrightarrow&|\longleftrightarrow|&
  \updownarrow&|\updownarrow|\cr
\Leftrightarrow&|\Leftrightarrow|&\Longleftrightarrow&|\Longleftrightarrow|&
  \Updownarrow&|\Updownarrow|\cr
\mapsto&|\mapsto|&\longmapsto&|\longmapsto|&
  \nearrow&|\nearrow|\cr
\hookleftarrow&|\hookleftarrow|&\hookrightarrow&|\hookrightarrow|&
  \searrow&|\searrow|\cr
\leftharpoonup&|\leftharpoonup|&\rightharpoonup&|\rightharpoonup|&
  \swarrow&|\swarrow|\cr
\leftharpoondown&|\leftharpoondown|&\rightharpoondown&|\rightharpoondown|&
  \nwarrow&|\nwarrow|\cr
\rightleftharpoons&|\rightleftharpoons|\cr
\enddisplay
Up and down arrows will grow larger, like delimiters (see Chapter~17).
To put symbols over left and right arrows, plain \TeX\ provides
a ^|\buildrel| macro: You type |\buildrel|\<superscript>^|\over|\<relation>,
and the superscript is placed on top of the relation just as limits are
placed over large operators. For example,
\begindisplay
\hbox to 30pt{$\buildrel \alpha\beta \over \longrightarrow$\hfil}%
|\buildrel \alpha\beta \over \longrightarrow|\cr
\hbox to 30pt{$\buildrel \rm def \over =$\hfil}%
|\buildrel \rm def \over =|\cr
\enddisplay
\ (In this context, `|\over|' does not define a fraction.)

\subsection Openings. The following ^{left delimiters} are available,
besides `(':
^^|\lbrack| ^^|\lbrace| ^^|\langle| ^^|\lfloor| ^^|\lceil|
\beginsymbols
\lbrack&|\lbrack|&\lfloor&|\lfloor|&\lceil&|\lceil|\cr
\lbrace&|\lbrace|&\langle&|\langle|\cr
\endsymbols
You can also type simply `|[|' to get |\lbrack|. All of these will grow if
you prefix them by ^|\bigl|, ^|\Bigl|, ^|\biggl|, ^|\Biggl|, or ^|\left|.
Chapter~17 also mentions ^|\lgroup| and ^|\lmoustache|, which are
available in sizes greater than |\big|. If you need more delimiters,
the following combinations work reasonably well in the normal text size:
\beginsymbols
\lbrack\!\lbrack&|\lbrack\!\lbrack|&
\langle\!\langle&|\langle\!\langle|&
(\!(&|(\!(|\cr
\endsymbols

\subsection Closings. The corresponding right delimiters are present too:
^^|\rbrack| ^^|\rbrace| ^^|\rangle| ^^|\rfloor| ^^|\rceil|
\beginsymbols
\rbrack&|\rbrack|&\rfloor&|\rfloor|&\rceil&|\rceil|\cr
\rbrace&|\rbrace|&\rangle&|\rangle|\cr
\endsymbols
Everything that works for openings works also for closings, but reversed.
\goodbreak

\subsection Punctuation. \TeX\ puts a thin space after commas and semicolons
that appear in mathematical formulas, and it does the same for a colon
that is called ^|\colon|. \ (Otherwise a ^{colon} is considered to be a
relation, as in `$x:=y$' and `$a:b::c:d$', which you type by saying
`|$x:=y$|' and `|$a:b::c:d$|'.) \ Examples of\/ |\colon| are
\begindisplay
\hbox to 115pt{$f\colon A\rightarrow B$\hfil}\enspace
|$f\colon A\rightarrow B$|\cr
\hbox to 115pt{$L(a,b;c\colon x,y;z)$\hfil}\enspace
|$L(a,b;c\colon x,y;z)$|\cr
\enddisplay
Plain \TeX\ also defines ^|\ldotp| and ^|\cdotp| to be `.'\ and `$\cdot$'
with the spacing of commas and semicolons. These symbols don't
occur directly in formulas, but they are useful in the definition of\/
^|\ldots| and ^|\cdots|.

\subsection Alternate names. If you don't like plain \TeX's name for
some math symbol---for example, if there's another name that looks
better or that you can remember more easily---the remedy is simple:
You just say, e.g., `|\let\cupcap=\asymp|'. Then you can type
`|f(n)\cupcap n|' instead of `|f(n)\asymp n|'.

Some symbols have alternate names that are so commonly used that plain
\TeX\ provides two or more equivalent control sequences:
^^|\ne| ^^|\neq| ^^|\le| ^^|\ge| ^^|\to| ^^|\gets| ^^|\owns| ^^|\land|
^^|\lor| ^^|\lnot| ^^|\vert|
$$\halign{\indent$\hfil#\hfil$\enspace&#\hfil\enspace&
  (same as #)\hfil\cr
\ne&|\ne| or |\neq|&|\not=|\cr
\le&|\le|&|\leq|\cr
\ge&|\ge|&|\geq|\cr
\{&|\{|&|\lbrace|\cr
\}&|\}|&|\rbrace|\cr
\to&|\to|&|\rightarrow|\cr
\gets&|\gets|&|\leftarrow|\cr
\owns&|\owns|&|\ni|\cr
\land&|\land|&|\wedge|\cr
\lor&|\lor|&|\vee|\cr
\lnot&|\lnot|&|\neg|\cr
\vert&|\vert|&\|\cr
\Vert&|\Vert|&|\|\|\cr
}$$
There's also ^|\iff| ($\iff$), which is just like |\Longleftrightarrow|
except that it puts an extra thick space at each side.

\subsection Non-math symbols. Plain \TeX\ makes four special symbols
available outside of math mode, although the characters themselves
are actually typeset from the math symbols font:
$$\halign{\indent$\hfil#\hfil$\enspace&#\hfil\enspace\cr
\S&|\S|\cr
\P&|\P|\cr
\dag&|\dag|\cr
\ddag&|\ddag|\cr
\phantom{\gets}\cr\noalign{\kern-\baselineskip}
}$$
These control sequences do not act like ordinary math symbols; they don't
change their size when they appear in subscripts or superscripts, and you
must say, e.g., |$x^{\P}$| instead of |$x^\P$| when you use them in
formulas. However, the ^|\dag| and ^|\ddag| symbols are available in
math mode under the names ^|\dagger| and ^|\ddagger|. It would be easy
to define mathematical equivalents of\/ ^|\S| and ^|\P|, if~these symbols
suddenly caught a mathematician's fancy.

\endchapter

Seek not for fresher founts afar,
Just drop your bucket where you are.
\author SAM WALTER ^{FOSS}, {\sl Back Country Poems\/} (1892)

\bigskip

No one compositor will have all the signs and symbols available.
The number of special signs and symbols is almost limitless,
with new ones being introduced all the time.
\author UNIVERSITY OF ^{CHICAGO} PRESS, {\sl Manual of Style\/} (1969) % p296

\eject
\beginchapter Appendix G. Generating Boxes\\from Formulas

\ninepoint
People who define new math fonts and/or macros sometimes need to know
exactly how \TeX\ manipulates the constituents of formulas. The purpose
of this appendix is to explain the precise positioning rules by which
\TeX\ converts a math list into a horizontal list. \ (It is a good idea
to review the introduction to ^{math lists} in Chapter~17 before
reading further; ``double dangerous bends'' are implied throughout
this appendix.)

\TeX\ relies on lots of parameters when it typesets formulas, and you have
the option of changing any or all of them. But of course you will want to
know what each parameter means, before you change it. Therefore each rule
below is numbered, and a table appears at the end to show which rules
depend on which parameters.

The most important parameters appear in the ^{symbol fonts} (family~2) and
the ^{extension fonts} (family~3). \TeX\ will not typeset a formula unless
^|\textfont||2|, ^|\scriptfont||2|, and ^|\scriptscriptfont||2| each contain
at least 22~^|\fontdimen| parameters. For brevity we shall call these parameters
$\sigma_1$ to $\sigma_{22}$, where the parameter is taken from
|\textfont2| if the current style is display or text ($D$ or $D'$ or $T$
or $T'$), from |\scriptfont2| if the current style is $S$ or $S'$, and from
|\scriptscriptfont2| otherwise. Similarly, the three fonts in family~3
must each have at least 13~|\fontdimen| parameters, and we will denote
them by $\xi_1$ to $\xi_{13}$. The notation $\xi_9$, for example, stands for
the ninth parameter of\/ |\scriptfont3|, if \TeX\ is typesetting something
in |\scriptstyle|.

A math list is a sequence of items of the various kinds listed in Chapter~17,
and \TeX\ typesets a formula by converting a math list to a horizontal
list. When such typesetting begins, \TeX\ has two other pieces of
information in addition to the math list itself. \ (a)~The starting style
tells what style should be used for the math list, unless another style
is specified by a style item. For example, the starting style for a
displayed formula is $D$, but for an equation in the text or an equation
number it is $T$; and for a subformula it can be any one of the eight
^{styles} defined in Chapter~17. We shall use $C$ to stand for the current
style, and we shall say that the math list is being typeset in style~$C$.
\ (b)~The typesetting is done either with or without penalties. Formulas
in the text of a paragraph are converted to horizontal lists in which
additional penalty items are inserted after binary operations and relations,
in order to aid in line breaking. Such penalties are not inserted in
other cases, because they would serve no useful function.

The eight styles are considered to be $D>D'>T>T'>S>S'>\it SS>SS'$, in
decreasing order. Thus, $C\le S$ means that the current style is $S$,
$S'$, $\it SS$, or~$\it SS'$.  Style $C'$ means the current style with a
prime added if one isn't there; for example, we have $C'=T'$ if and only
if $C=T$ or $C=T'$.  Style~$C\mathord\uparrow$ is the superscript style
for $C$; this means style~$S$ if $C$~is $D$ or~$T$, style~$S'$ if $C$~is
$D'$ or $T'$, style~$\it SS$ if $C$~is $S$ or $\it SS$, and style~$\it
SS'$ if $C$~is $S'$ or $\it SS'$.  Finally, style~$C\mathord\downarrow$ is
the subscript style, which is~$(C\mathord\uparrow)'$.

Chapter 17 stated that the most important components of math lists are
called atoms, and that each atom has three fields called its nucleus,
subscript, and superscript.  We frequently need to execute a subroutine
called ``Set box~$x$ to the so-and-so field in style such-and-such.'' This
means (a)~if the specified field is empty, $x$~is set equal to a null box;
(b)~if the field contains a symbol, $x$~is set to an hbox containing that
symbol in the appropriate size, and the ^{italic correction} for the
character is included in the width of the box; (c)~if the field contains a
math list or horizontal list, $x$~is set to an hbox containing the result
of typesetting that list with the specified starting style. In case~(c),
the glue is set with no stretching or shrinking, and an additional level
of hboxing is omitted if it turns out to be redundant.

Another subroutine sets box $x$ to a specified variable ^{delimiter},
having a specified minimum height plus depth.  This means that a search is
conducted as follows: The delimiter is defined by two symbols, a ``small
character''~$a$ in family~$f$ and a ``large character''~$b$ in family~$g$.
The search looks first at character $a$ in scriptscriptfont~$f$, if $C\le
\it SS$; then it looks at $a$ in scriptfont~$f$, if $C\le S$; then it looks at
$a$ in textfont~$f$.  If nothing suitable is found from $a$ and~$f$, the
larger alternative $b$ and $g$ is examined in the same way.  Either
$(a,f)$ or $(b,g)$ may be $(0,0)$, which means that the corresponding part
of the search is to be bypassed.  When looking at a character in a
font, the search stops immediately if that character has sufficient height
plus depth, or if the character is ^{extensible}; furthermore, if the
character does not stop the search, and if it has a ^{successor} in the
font, the successor is looked at next. \ (See the \MF\ manual or the
^^{METAFONT}
system documentation of |tfm| files for further information about
successors and extensible characters.) \ If the search runs all the way to
completion without finding a suitable character, the one with greatest
height plus depth is chosen. If no characters at all were found (either
because $a=f=b=g=0$ or because the characters did not exist in the fonts),
$x$~is set to an empty box whose width is ^|\nulldelimiterspace|.  If an
extensible character was found, $x$~is set to a vbox containing enough
pieces to build up a character of sufficient size; the height of this vbox
is the height of the topmost piece, and the width is the width of the
repeatable piece. ^^{built-up characters} Otherwise
$x$~is set to an hbox containing the character that was found; the italic
correction of the character is included in the width of this box.

There's also a subroutine that ``reboxes'' a given box to a given~width.
If the box doesn't already have the desired width, \TeX\ unpackages it
(unless it was a vbox), then adds a kern for an italic correction if one was
implied, and inserts ^|\hss| glue at both left and right; the resulting
horizontal list is packaged into an hbox.  This process is used, for
example, to give a common width to the numerator and denominator of a
fraction; it centers whichever is smaller, unless infinite glue is present
in addition to the newly added |\hss|.

If $x$ is a box, we shall use the abbreviations $h(x)$, $d(x)$, and $w(x)$ for
its height, depth, and width, respectively.

Here now are the rules for typesetting a given math list in starting
style~$C$. The process applies from left to right, translating each
item in turn. Two passes are made over the list;
most of the work is done by the first pass, which compiles individual
translations of the math items. We shall consider this part of the
task first:

\def\rule#1.{\smallskip\textindent{\bf#1.}\ignorespaces}
\rule 1. If the current item is a rule or discretionary or penalty or
``whatsit'' or boundary item, simply leave it unchanged and move to the
next item.

\rule 2. If the current item is glue or a kern, translate it as follows:
If it is glue from ^|\nonscript|, check if the immediately following item
is glue or a kern; and if so, remove that item if $C\le S$. Otherwise, if the
current item is from ^|\mskip| or ^|\mkern|, convert from |mu| to absolute
units by multiplying each finite dimension by ${1\over18}\sigma_6$. Then
move on to the next item.

\rule 3. If the current item is a style change, set $C$ to the specified
style and move on to the next item.

\rule 4. If the current item is a four-way choice, it contains four math
lists for the four main styles. Replace it by the math list that corresponds
to the current style $C$, then move to the first unprocessed item.

\rule 5. If the current item is a Bin atom, and if this was the first atom
in the list, or if the most recent previous atom was Bin, Op, Rel, Open,
or Punct, change the current Bin to Ord and continue with Rule~14.
Otherwise continue with Rule~17.

\rule 6. If the current item is a Rel or Close or Punct atom, and if the most
recent previous atom was Bin, change that previous Bin to Ord. Continue
with Rule~17.

\rule 7. If the current item is an Open or Inner atom, go directly to Rule~17.

\rule 8. If the current item is a Vcent atom (from ^|\vcenter|), let its
nucleus be a vbox of height-plus-depth~$v$. Change the height to
${1\over2}v+a$ and the depth to ${1\over2}v-a$, where $a$~is the
axis height, $\sigma_{22}$. Change this atom to type Ord and continue
with Rule~17.

\rule 9. If the current item is an Over atom (from ^|\overline|), set
box~$x$ to the nucleus in style $C'$. Then replace the nucleus by a vbox
containing kern~$\theta$, hrule of~height~$\theta$, kern~$3\theta$,
and box~$x$, from top to bottom, where $\theta=\xi_8$ is the ^{default
rule thickness}. \ (This puts a rule over the nucleus, with $3\theta$ clearance,
and with $\theta$~units of extra white space assumed to be present above
the rule.) \ Continue with Rule~16.

\rule 10. If the current item is an Under atom (from ^|\underline|), set
box~$x$ to the nucleus in style $C$. Then replace the nucleus by a vtop
made from box~$x$, kern~$3\theta$, and hrule of~height~$\theta$,
where $\theta=\xi_8$ is the default rule thickness; and add $\theta$ to the
depth of the box. \ (This puts a rule under the nucleus, with $3\theta$
clearance, and with $\theta$~units of extra white space assumed to be
present below the rule.) \ Continue with Rule~16.

\rule 11. If the current item is a Rad atom (from ^|\radical|, e.g.,
^|\sqrt|), set box~$x$ to the nucleus in style $C'$. Let $\theta=\xi_8$;
and let $\varphi=\sigma_5$ if $C>T$, otherwise $\varphi=\theta$.
Set $\psi=\theta+{1\over4}\vert\varphi\vert$; this is the minimum
clearance that will be allowed between box~$x$ and the rule that will go
above it. Set box~$y$ to a variable delimiter for this radical atom, having
height plus depth $h(x)+d(x)+\psi+\theta$ or more. Then set $\theta\leftarrow
h(y)$; this is the thickness of the rule to be used in the radical
construction. \ (Note that the font designer specifies the thickness of
the rule by making it the height of the radical character; the baseline
of the character should be precisely at the bottom of the rule.) \
If $d(y)>h(x)+d(x)+\psi$, increase $\psi$ by half of the excess; i.e.,
set $\psi\leftarrow{1\over2}\bigl(\psi+d(y)-h(x)-d(x)\bigr)$. Construct
a vbox consisting of kern~$\theta$, hrule of~height~$\theta$, kern~$\psi$,
and box~$x$, from top to bottom. The nucleus of the radical atom is now
replaced by box~$y$ raised by $h(x)+\psi$, followed by the new vbox.
Continue with Rule~16.

\rule 12. If the current item is an Acc atom (from ^|\mathaccent|), just
go to Rule~16 if the accent character doesn't exist in the current size.
Otherwise set box~$x$ to the nucleus in style $C'$, and set $u$ to the
width of this box.  If the nucleus is not a single character, let $s=0$;
otherwise set $s$ to the kern amount for the nucleus followed by the
^|\skewchar| of its font.  If the accent character has a successor in its
font whose width is $\le u$, change it to the successor and repeat this
sentence.  Now set $\delta\leftarrow \min\bigl(h(x),\chi\bigr)$, where
$\chi$ is |\fontdimen5| (the ^{x-height}) in the accent font.  If the
nucleus is a single character, replace box~$x$ by a box containing the
nucleus together with the superscript and subscript of the Acc atom, in
style~$C$, and make the sub/superscripts of the Acc atom empty; also
increase $\delta$ by the difference between the new and old values
of~$h(x)$. Put the accent into a new box~$y$, including the italic
correction.  Let~$z$ be a vbox consisting of: box~$y$ moved right
$s+{1\over2}\bigl(u-w(y)\bigr)$, kern $-\delta$, and box~$x$. If
$h(z)<h(x)$, add a kern of $h(x)-h(z)$ above box~$y$ and set $h(z)\gets
h(x)$. Finally set $w(z)\gets w(x)$, replace the nucleus of the Acc atom
by box~$z$, and continue with Rule~16.

\rule 13. If the current item is an Op atom, mark this atom as having
limits if it has been marked with ^|\limits|, or if it has been marked
with ^|\displaylimits| and $C>T$.  If the nucleus is not a symbol, set
$\delta\leftarrow0$ and go to Rule~13a. Otherwise if $C>T$ and if the
nucleus symbol has a successor in its font, move to the successor. \ (This
is where operators like $\sum$ and $\int$ change to a larger size in
display styles.) \ Put the symbol into a new box~$x$, in the current size,
and set $\delta$ to the italic correction for the character; include $\delta$
in the width of box~$x$ if and only if limits are to be set or
there is no subscript. Shift box~$x$ down by ${1\over2}\bigl(h(x)-d(x)\bigr)
-a$, where $a=\sigma_{22}$, so that the operator character is centered
vertically on the axis; this shifted box becomes the nucleus of the Op atom.

\rule 13a. If limits are not to be typeset for this Op atom, go to
Rule~17; otherwise the limits are attached as follows:
Set box~$x$ to the superscript field in style $C\mathord\uparrow$; set box~$y$
to the nucleus field in style~$C$; and set box~$z$ to the subscript field
in style $C\mathord\downarrow$. Rebox all three of these boxes to width
$\max\bigl(w(x),w(y),w(z)\bigr)$.  If the superscript field was not empty,
attach box~$x$ above box~$y$, separated by a kern of size
$\max\bigl(\xi_9,\xi_{11}-d(x)\bigr)$, and shift box~$x$ right by
${1\over2}\delta$; also put a kern of size~$\xi_{13}$ above box~$x$.
If the subscript field was not empty,
attach box~$z$ below box~$y$, separated by a kern of size
$\max\bigl(\xi_{10},\xi_{12}-h(z)\bigr)$, and shift box~$z$ left by
${1\over2}\delta$; also put a kern of size~$\xi_{13}$ below box~$z$.
The resulting vbox becomes the nucleus of the current Op atom; move to
the next item.

\rule 14. If the current item is an Ord atom, go directly to Rule~17 unless
all of the following are true: The nucleus is a symbol; the subscript
and superscript are both empty; the very next item in the math list is an
atom of type Ord, Op, Bin, Rel, Open, Close, or Punct; and the nucleus of the
next item is a symbol whose family is the same as the family in the present
Ord atom. In such cases the present symbol is marked as a text symbol.
If the font information shows a ligature between this symbol and the
following one, using the specified family and the current size, then
insert the ligature character and continue as specified by the font;
in this process, two characters may collapse into a single Ord
text symbol, and/or new Ord text characters may appear. If the font information
shows a kern between the current symbol and the next, insert a kern item
following the current atom.
As soon as an Ord atom has been fully processed for ligatures and kerns,
go to Rule~17.

\rule 15. If the current item is a generalized fraction (and it had better
be, because that's the only possibility left if Rules 1--14 don't apply),
let $\theta$ be the thickness of the bar line and let $(\lambda,\rho)$ be
the left and right delimiters. \ If this fraction was generated by
^|\over| or ^|\overwithdelims|, then $\theta=\xi_8$; if it was generated by
^|\atop| or ^|\atopwithdelims|, $\theta=0$; otherwise it was generated by
^|\above| or ^|\abovewithdelims|, and a specific value of~$\theta$ was
given at that time. The values of $\lambda$ and $\rho$ are null unless
the fraction is ``with delims.''

\rule 15a. Put the numerator into box~$x$, using style $T$ or $T'$ if
$C$~is $D$ or $D'$, otherwise using style $C\mathord\uparrow$.
Put the denominator into box~$z$, using style $T'$ if
$C>T$, otherwise using $C\mathord\downarrow$.
If $w(x)<w(z)$, rebox $x$ to width~$w(z)$;
if $w(z)<w(x)$, rebox $z$ to width~$w(x)$.

\rule 15b. If $C>T$, set $u\leftarrow\sigma_8$ and $v\leftarrow\sigma_{11}$.
Otherwise set $u\leftarrow\sigma_9$ or $\sigma_{10}$, according as $\theta\ne0$
or $\theta=0$, and set $v\leftarrow\sigma_{12}$. \ (The fraction will be
typeset with its numerator shifted up by an amount~$u$ with respect to
the current baseline, and with the denominator shifted down by~$v$,
unless the boxes are unusually large.)

\rule 15c. If $\theta=0$ (|\atop|), the numerator and denominator are
combined as follows: Set $\varphi\leftarrow7\xi_8$ or
$3\xi_8$, according as $C>T$ or $C\le T$; $\varphi$ is the minimum
clearance that will be tolerated between numerator and denominator.
Let $\psi=\bigl(u-d(x)\bigr)-\bigl(h(z)-v\bigr)$ be the actual clearance
that would be obtained with the current values of $u$ and~$v$; if
$\psi<\varphi$, add ${1\over2}(\varphi-\psi)$ to both $u$ and~$v$.
Then construct a vbox of height $h(x)+u$ and depth $d(z)+v$, consisting
of box~$x$ followed by an appropriate kern followed by box~$z$.

\rule 15d. If $\theta\ne0$ (|\over|), the numerator and denominator are
combined as follows: Set $\varphi\leftarrow3\theta$ or $\theta$, according
as $C>T$ or $C\le T$; $\varphi$ is the minimum clearance that will be
tolerated between numerator or denominator and the bar line.  Let
$a=\sigma_{22}$ be the current axis height; the middle of the bar line
will be placed at this height.  If
$\bigl(u-d(x)\bigr)-(a+{1\over2}\theta)<\varphi$, increase~$u$ by the
difference between these quantities; and if $(a-{1\over2}\theta)-
\bigl(h(z)-v\bigr)<\varphi$, increase~$v$ by the difference. Finally
construct a vbox of height $h(x)+u$ and depth $d(z)+v$, consisting of
box~$x$ followed by a kern followed by an hrule of height~$\theta$
followed by another kern followed by box~$z$, where the kerns are figured
so that the bottom of the hrule occurs at $a-{1\over2}\theta$ above the
baseline.

\rule 15e. Enclose the vbox that was constructed in Rule 15c or 15d by
delimiters $(\lambda,\rho)$
whose height plus depth is at least $\sigma_{20}$, if $C>T$, and at
least $\sigma_{21}$ otherwise. Shift the delimiters up or down so that they are
vertically centered with respect to the axis. Replace the generalized
fraction by an Ord atom whose nucleus is the resulting sequence of three boxes
($\lambda$, vbox, $\rho$). Go to rule~19.

\bigbreak\noindent
Rules 1--15 account for the preliminary processing of math list items;
but we still haven't specified how subscripts and superscripts are to be
typeset. Therefore some of those rules lead to the following post-process:

\rule 16. Change the current item to an Ord atom, and continue with Rule~17.

\rule 17. If the nucleus of the current item is a math list, replace it by
a box obtained by typesetting that list in the current style.  Then if the
nucleus is not simply a symbol, go on to Rule~18.  Otherwise we are in the
common case that a math symbol is to be translated to its horizontal-list
equivalent: Convert the symbol to a character box for the specified family
in the current size. If the symbol was not marked by Rule~14 above as a
text symbol, or if\/ |\fontdimen| parameter number~2 of its font is zero, set
$\delta$ to the italic correction; otherwise set $\delta$ to zero. If
$\delta$ is nonzero and if the subscript field of the current atom is
empty, insert a kern of width~$\delta$ after the character box, and set
$\delta$ to zero. Continue with Rule~18.

\rule 18. (The remaining task for the current atom is to attach a possible
subscript and superscript.) \ If both subscript and superscript fields
are empty, move to the next item. Otherwise continue with the following
subrules:

\rule 18a. If the translation of the nucleus is a character box, possibly
followed by a kern, set $u$ and~$v$ equal to zero; otherwise set
$u\leftarrow h-q$ and $v\leftarrow d+r$, where $h$ and~$d$ are the height
and depth of the translated nucleus, and where $q$ and~$r$ are the values
of $\sigma_{18}$ and $\sigma_{19}$ in the font corresponding to styles
$C\mathord\uparrow$ and~$C\mathord\downarrow$.  \ (The quantities $u$
and~$v$ represent minimum amounts by which the superscript and subscript
will be shifted up and down; these preliminary values of $u$ and~$v$ may
be increased later.)

\rule 18b. If the superscript field is empty (so that there is a subscript
only), set box~$x$ to the subscript in style~$C\mathord\downarrow$, and add
^|\scriptspace| to $w(x)$. Append this box to the translation of the
current item, shifting it down by $\max\bigl(v,\sigma_{16},h(x)-{4\over5}
\vert\sigma_5\vert\bigr)$, and move to the next item. \ (The idea is
to make sure that the subscript is shifted by at least~$v$ and by at
least $\sigma_{16}$; furthermore, the top of the subscript should not extend
above $4\over5$~of the current x-height.)

\rule 18c. Set box~$x$ to the superscript field in style~$C\mathord\uparrow$,
and add |\scriptspace| to~$w(x)$. Then set $u\leftarrow\max\bigl(u,p,
d(x)+{1\over4}\vert\sigma_5\vert\bigr)$, where $p=\sigma_{13}$ if $C=D$,
$p=\sigma_{15}$ if $C=C'$, and $p=\sigma_{14}$ otherwise; this gives a
tentative position for the superscript.

\rule 18d. If the subscript field is empty (so that there is a
superscript only), append box~$x$ to the translation of the current atom,
shifting it up by $u$, and move to the next item. Otherwise (i.e.,
both subscript and superscript are present), set box~$y$ to the
subscript in style~$C\mathord\downarrow$, add |\scriptspace| to~$w(y)$,
and set $v\leftarrow\max(v,\sigma_{17})$.

\rule 18e. (The remaining task is to position a joint
subscript/superscript combination.) \ Let $\theta=\xi_8$ be the default
rule thickness.  If $\bigl(u-d(x)\bigr)-\bigl(h(y)-v\bigr)\ge4\theta$, go
to Rule~18f. \ (This means that the white space between subscript and
superscript is at least $4\theta$.) \ Otherwise reset $v$ so that
$\bigl(u-d(x)\bigr)-\bigl(h(y)-v\bigr)=4\theta$. Let $\psi={4\over5}
\vert\sigma_5\vert-\bigl(u-d(x)\bigr)$. If $\psi>0$, increase $u$
by~$\psi$ and decrease $v$ by~$\psi$. \ (This means that the bottom of the
superscript will be at least as high above the baseline as $4\over5$ of
the x-height.)

\rule 18f. Finally, let $\delta$ be zero unless it was set to a nonzero
value by Rules 13 or 17. \ (This is the amount of horizontal displacement
between subscript and superscript.) Make a vbox of height $h(x)+u$ and
depth $d(y)+v$, consisting of box~$x$ shifted right by~$\delta$, followed
by an appropriate kern, followed by box~$y$. Append this vbox to the
translation of the current item and move to the next.

\bigbreak\noindent
After the entire math list has been processed by Rules 1--18, \TeX\ looks
at the last atom (if there was one), and changes its type from
Bin to Ord (if it was of type Bin). Then the following rule is performed:

\rule 19. If the math list begins and ends with boundary items, compute
the maximum height~$h$ and depth~$d$ of the boxes in the translation of
the math list that was made on the first pass, taking into account the
fact that some boxes may be raised or lowered. Let $a=\sigma_{22}$ be the
axis height, and let $\delta=\max(h-a,d+a)$ be the amount by which the
formula extends away from the axis. Replace the boundary items by
delimiters whose height plus depth is at least $\max(\lfloor\delta/500\rfloor
f,2\delta-l)$, where $f$ is the ^|\delimiterfactor|
and $l$ is the ^|\delimitershortfall|.  Shift the delimiters up or down so
that they are vertically centered with respect to the axis. Change the left
boundary item to an Open atom and the right boundary item to
a Close atom. The entire resulting list now becomes the nucleus of an
Inner atom. \ (All of the calculations in this step are done with
$C$ equal to the starting style of the math list; style items in the
middle of the list do not affect the style of the right boundary item.)

\medbreak \rule 20. Rules 1--19 convert the math list into a sequence of
items in which the only remaining atoms are of types Ord, Op, Bin, Rel,
Open, Close, Punct, and Inner. After that conversion is complete, a second pass
is made through the entire list, replacing all of the atoms by the boxes
and kerns in their translations. Furthermore, additional inter-element
spacing is inserted just before each atom except the first, based on the
type of that atom and the preceding one.  Inter-element spacing is defined
by the three parameters ^|\thinmuskip|, ^|\medmuskip|, and
^|\thickmuskip|; the |mu| units are converted to absolute units as in
Rule~2 above.  Chapter~18 has a chart that defines the inter-element
spacing, some of which is ^|\nonscript|, i.e., it is inserted only in
styles $>S$.  The list might also contain style items, which are removed
during the second pass; they are used to change the current style just as
in the first pass, so that both passes have the same value of~$C$ when
they work on any particular atom.

\rule 21. Besides the inter-element spacing, penalties are placed after
the translation of each atom of type Bin or Rel, if the math list was part
of a paragraph.  The penalty after a Bin is ^|\binoppenalty|, and the
penalty after a Rel is ^|\relpenalty|. However, the penalty is not
inserted after the final item in the entire list, or if it has a numeric
value~$\ge10000$, or if the very next item in the list is already a
penalty item, or after a Rel atom that is immediately followed by another
Rel atom.

\rule 22. After all of the preceding actions have been performed, the math
list has been totally converted to a horizontal list. If the result is
being inserted into a larger horizontal list, in horizontal mode or
restricted horizontal mode, it is enclosed by ``^{math-on}'' and
``^{math-off}'' items that each record the current value of\/
^|\mathsurround|.  Or if this list is a displayed formula, it
is processed further as explained in Chapter~19.

\medbreak
\def\\{\kern1pt}
\noindent{\bf Summary of parameter usage.} \ Here is the promised index
that refers to everything affected by the mysterious parameters
in the symbol fonts. Careful study of the rules allows you to get the
best results by appropriately setting the parameters for new fonts
that you may wish to use in mathematical typesetting. Each font parameter
has an external name that is used in supporting software packages;
for example, $\sigma_{14}$ is generally referred to as `sup2' and $\xi_8$ as
`default\_\\ rule\_thickness'. These external names are indicated in the table.
$$\vbox{
\halign to\hsize{$#\hfil$\tabskip10pt plus 10pt minus 3pt&
#\hfil&$\rm#\hfil$\quad&
$#\hfil$&#\hfil&$\rm#\hfil$\tabskip0pt\cr
\omit\span\omit\it Parameter&\omit\it Used in\hfil&
\omit\span\omit\it Parameter&\omit\it Used in\hfil\cr
\noalign{\medskip\hrule\smallskip}
\sigma_5&x\_\\height&11, 18b, 18c, 18e&
\sigma_{18}&sup\_drop&18a\cr
\sigma_6&quad&2, 20&
\sigma_{19}&sub\_drop&18a\cr
\sigma_8&num1&15b&
\sigma_{20}&delim1&15e\cr
\sigma_9&num2&15b&
\sigma_{21}&delim2&15e\cr
\sigma_{10}&num3&15b&
\sigma_{22}&axis\_\\height&8, 13, 15d, 19\cr
\sigma_{11}&denom1&15b&
\xi_8&default\_\\rule\_thickness&9, 10, 11, 15, 15c, 18e\cr
\sigma_{12}&denom2&15b&
\xi_9&big\_op\_\\spacing1&13a\cr
\sigma_{13}&sup1&18c&
\xi_{10}&big\_op\_\\spacing2&13a\cr
\sigma_{14}&sup2&18c&
\xi_{11}&big\_op\_\\spacing3&13a\cr
\sigma_{15}&sup3&18c&
\xi_{12}&big\_op\_\\spacing4&13a\cr
\sigma_{16}&sub1&18b&
\xi_{13}&big\_op\_\\spacing5&13a\cr
\sigma_{17}&sub2&18d\cr
\noalign{\smallskip\hrule\medskip}}}$$
Besides the symbol and extension fonts (families 2 and~3), the
rules above also refer to parameters in arbitrary families:
Rule~17 uses |\fontdimen| parameter~2 (space) to determine whether to insert
an italic correction between adjacent letters, and Rule~12 uses parameter~5
(x\_\\height) to position an accent character. Several non-font parameters
also affect mathematical typesetting:
dimension parameters
 \hbox{|\delimitershortfall|} (Rule~19),
 \hbox{|\nulldelimiterspace|} (in
the construction of variable delimiters for Rules 11, 15e,~19),
 \hbox{|\mathsurround|} (Rule~22),
and
 \hbox{|\scriptspace|} (Rules 18bcd);
integer parameters
 \hbox{|\delimiterfactor|} (Rule~19),
 \hbox{|\binoppenalty|} (Rule~21),
and
 \hbox{|\relpenalty|} (Rule~21);
muglue parameters
 \hbox{|\thinmuskip|},
 \hbox{|\medmuskip|,} and
 \hbox{|\thickmuskip|}
(Rule~20).

\endchapter

Woe to the author who always wants to teach!\/
% Mais malheur a` l'auteur qui veut toujours instruire!
The secret of being a bore is to tell everything.
% Le secret d'ennuyer est celui de tout dire.
\author ^{VOLTAIRE}, {\sl De la Nature de l'Homme\/} (1737)
% 6th discours en vers sur l'homme

\bigskip

Very few Compositors are fond of Algebra,
and rather chuse to be employed upon plain work.
%\author PHILIP ^{LUCKOMBE}, {\sl The History and Art of Printing\/} (1770)
% page 472 ... but I discovered on 3 June 1998 that Luckombe plagiarized!
\author JOHN ^{SMITH}, {\sl The Printer's Grammar\/} (1755)
% page 308
\eject
\beginchapter Appendix H. Hyphenation

It's better to break a word with a hyphen than to stretch interword spaces
too much. Therefore \TeX\ tries to divide words into syllables when
there's no good alternative available.

But computers are notoriously bad at ^{hyphenation}. When the type\-setting of
newspapers began to be fully automated, ^{jokes} about ``the-rapists who
pre-ached on wee-knights'' soon began to circulate.

It's not hard to understand why machines have behaved poorly at this task,
because hyphenation is quite a difficult problem. For example, the word
`record' is supposed to be broken as `rec-ord' when it is a noun, but
`re-cord' when it is a verb. The word `hyphenation' itself is somewhat
exceptional; if `hy-phen-a-tion' is compared to similar words like
`con-cat-e-na-tion', it's not immediately clear why the `n' should
be attached to the `e' in one case but not the other. Examples like
`dem-on-stra-tion' vs.~`de-mon-stra-tive' show that the alteration of
two letters can actually affect hyphens that are nine positions away.

A good solution to the problem was discovered by Frank M. ^{Liang} during
1980--1982, and \TeX\ incorporates the new method. Liang's algorithm
works quickly and finds nearly all of the legitimate places to insert
hyphens; yet it makes few if any errors, and it takes up comparatively
little space in the computer. Moreover, the method is flexible enough
to be adapted to any language, and it can also be used to hyphenate
words in two languages simultaneously. Liang's Ph.D. thesis,
published by Stanford University's Department of Computer Science in~1983,
explains how to take a dictionary of hyphenated words and teach it to \TeX;
i.e., it explains how to compute tables by which \TeX\ will be able to
reconstruct most of the hyphens in the given dictionary, without error.

\def\\#1{$_{\kern\scriptspace#1}$}
\TeX\ hyphenates a given word by first looking for it in an ``^{exception
dictionary},'' which specifies the hyphen positions for words that deserve
special treatment. If the word isn't there, \TeX\ looks for {\sl^{patterns}\/}
in the word, and this is the key idea underlying Liang's method. Here's how
it works, using the word `hyphenation' as an example, when \TeX\ is
operating with the English-oriented patterns of plain \TeX\ format:
The given word is first extended by special markers at either end; in this
case we obtain
\begintt
.hyphenation.
\endtt
if `|.|'\ denotes the special marker. The extended word has subwords
\begintt
. h y p h e n a t i o n .
\endtt
of length one,
\begintt
.h hy yp ph he en na at ti io on n.
\endtt
of length two,
\begintt
.hy hyp yph phe hen ena nat ati tio ion on.
\endtt
of length three,
and so on. Each subword of length $k$ is a pattern that defines $k+1$ small
integer values relating to the desirability of hyphens in the positions
between and adjacent to its letters. We can show these values by attaching
them as subscripts; for example,
`{\tt\\0h\\0e\\2n\\0}' means that the values corresponding to the
subword `|hen|' are 0,~0, 2, and~0, where the~2 relates to hyphens
between the `|e|' and the `|n|'. The interletter values are entirely zero
for all subwords except those that match an entry in \TeX's current
``^{pattern dictionary}''; and in this case, only the subwords
\begindisplay
\tt\\0h\\0y$_3$p\\0h\\0 \\0h\\0e\\2n\\0 \\0h\\0e\\0n\\0a\\4\cr
\qquad\tt\\0h\\0e\\0n\\5a\\0t\\0 \\1n\\0a\\0 \\0n\\2a\\0t\\0
  \\1t\\0i\\0o\\0 \\2i\\0o\\0 \\0o\\2n\\0\cr
\enddisplay
occur as special patterns. \TeX\ now computes the maximum interletter
value that occurs at each subword touching each interletter position.
For example, between `|e|' and `|n|' there are five relevant values
in this case (2~from {\tt\\0h\\0e\\2n\\0},
0~from {\tt\\0h\\0e\\0n\\0a\\4},
0~from {\tt\\0h\\0e\\0n\\5a\\0t\\0},
1~from {\tt\\1n\\0a\\0},
and 0~from {\tt\\0n\\2a\\0t\\0}); the maximum of these is~2.
The result of all the maximizations is
\begindisplay
\tt.\\0h\\0y$_3$p\\0h\\0e\\2n\\5a\\4t\\2i\\0o\\2n\\0.
\enddisplay
Now comes the final step: A hyphen is considered to be acceptable
between two letters if the associated interletter value is {\sl odd}.
Thus, two potential breakpoints have been found: `|hy-phen-ation|'.
Similarly, the word `|concatenation|' contains the patterns
\begindisplay
\tt\\0o\\2n\\0
\\0o\\0n\\1c\\0
\\1c\\0a\\0
\\1n\\0a\\0
\\0n\\2a\\0t\\0
\\1t\\0i\\0o\\0
\\2i\\0o\\0
\\0o\\2n\\0
\enddisplay
and this yields `{\tt\\0c\\0o\\2n\\1c\\0a\\0t\\0e\\1n\\2a\\1t\\2i\\0o\\2n\\0}',
i.e., `|con-cate-na-tion|'.

Let's try a 34-letter word:
`|supercalifragilisticexpialidocious|' matches the plain \TeX\ patterns
\begindisplay
\tt u\\1pe r\\1c \\1ca al\\1i ag\\1i gil\\4 il\\1i il\\4ist is\\1ti st\\2i\cr
\qquad\tt s\\1tic
  \\1exp x\\3p pi\\3a \\2i\\1a i\\2al \\2id \\1do \\1ci \\2io ou\\2 \\2us\cr
\enddisplay
(where subscripts that aren't shown are zero), and this yields
$$\centerline{%
\tt.\\0s\\0u\\1p\\0e\\0r\\1c\\0a\\0l\\1i\\0f\\0r\\0a\\0g\\1i\\0l\\4i%
\\0s\\1t\\2i\\0c\\1e\\0x\\3p\\2i\\3a\\0l\\2i\\1d\\0o\\1c\\2i\\0o\\2u\\2s\\0.}$$
The \stretch resulting \stretch hyphens \stretch
`|su-|\stretch|per-|\stretch|cal-|\stretch|ifrag-|\stretch|ilis-|\stretch
|tic-|\stretch|ex-|\stretch|pi-|\stretch|ali-|\stretch|do-|\stretch|cious|'
agree with Random House's {\sl Unabridged Dictionary\/} (which also
shows a few more: `|su-per-cal-i-frag-i-lis-tic-ex-pi-al-i-do-cious|').

Plain \TeX\ loads exactly 4447 patterns into \TeX's memory, beginning with
`{\tt\\0.\\0a\\0c\\0h\\4}' and ending with `{\tt\\4z\\1z\\2}' and
`{\tt\\0z\\4z\\0y$_0$}'. The interletter values in these patterns are all
between 0 and~5; a large odd value like the 5 in `{\tt\\0h\\5e\\0l\\0o\\0}'
forces desirable hyphen points in words like `|bach-e-lor|' and
`|ech-e-lon|', while a large even value like the 4 in `{\tt\\0h\\0a\\0c\\0h\\4}'
suppresses undesirable hyphens in words like `|tooth-aches|'. Liang
derived these patterns by preparing a special version of {\sl Webster's
Pocket Dictionary\/} (Merriam, 1966) that contains about 50,000 words including
derived forms. Then he checked a preliminary set of patterns obtained from
this data against an up-to-date hyphenation dictionary of about
115,000 words obtained from a publisher; errors found in this run
led to the addition of about 1000 words like |camp-fire|,
|Af-ghan-i-stan|, and |bio-rhythm| to the pocket dictionary list. He
weighted a few thousand common words more heavily so that they would be more
likely to be hyphenated; as a result, the patterns of plain \TeX\ guarantee
complete hyphenation of the 700 or so most common words of English,
as well as common technical words like |al-go-rithm|. These patterns
find 89.3\% of the hyphens in Liang's dictionary as a whole, and
they insert no hyphens that are not present.

Patterns derived from the common words of a language tend to work well
on uncommon or newly coined words that are not in the original dictionary.
For example, Liang's patterns find a correct subset of the hyphens in the
word that all of today's unabridged dictionaries agree is the longest
in English, namely
\begintt
pneu-monoul-tra-mi-cro-scop-ic-sil-i-co-vol-canoco-nio-sis.
\endtt
They even do fairly well on words from other languages that aren't
too distant from English; for example, the pseudo-^{German} utterances
of Mark ^{Twain}'s {\sl Connecticut Yankee\/} come out with only nine or
ten bad hyphens: % see Chapter 23 of Twain's novel
\begintt
Con-stanti-nop-o-li-tanis-cher-
  dudel-sack-spfeifen-mach-ers-ge-sellschafft;
Ni-hilis-ten-dy-na-mitthe-
  aterkaestchensspren-gungsat-ten-taetsver-suchun-gen;
Transvaal-trup-pen-tropen-trans-port-
  tram-pelth-iertreib-er-trau-ungsthrae-nen-tra-goedie;
Mekka-musel-man-nen-massen-menchen-
  mo-er-der-mohren-mut-ter-mar-mor-mon-u-menten-macher.
\endtt
% Note: I've faithfully copied Twain's errors, like "menchen" for "menschen"
% and "attentaets" for "attentats"; also, TeX might suppress the final
% hyphens in ultralong words, but I'm showing what the patterns say
But when plain \TeX\ is tried on the name of a famous ^{Welsh} village,
^^{Llanfair P. G.}
\begintt
Llan-fair-p-wll-gwyn-gyll-gogerych-
  wyrn-drob-wl-l-l-lan-tysil-i-o-gogogoch,
\endtt
linguistic differences became quite evident, since the correct hyphens are
\begintt
Llan-fair-pwll-gwyn-gyll-go-ger-y-
  chwyrn-dro-bwll-llan-ty-sil-i-o-go-go-goch.
\endtt
Appropriate pattern values for other languages can be derived by applying
Liang's method to suitable dictionaries of hyphen points.

^{Dictionaries} of English do not always agree on where syllable boundaries
occur. For example, the {\sl American Heritage Dictionary\/} says
`|in-de-pend-ent|' while {\sl Webster's\/} says `|in-de-pen-dent|'.
Plain \TeX\ generally follows Webster except in a few cases where other
authorities seem preferable.

\begingroup
\hyphenpenalty=-1000 \pretolerance=-1 \tolerance=1000
\doublehyphendemerits=-100000 \finalhyphendemerits=-100000
[From here to the end of this appendix, \TeX\ will be typesetting with
\begindisplay
^|\hyphenpenalty||=-1000   |^|\pretolerance||=-1   |^|\tolerance||=1000|\cr
^|\doublehyphendemerits||=-100000  |^|\finalhyphendemerits||=-100000|\cr
\enddisplay
so that hyphens will be inserted much more often than usual.]

\looseness-1
The fact that plain \TeX\ finds only 90\% of the permissible hyphen points
in a large dictionary is, of course, no cause for alarm. When word
frequency is taken into account, the probability rises to well over 95\%.
Since \TeX's line-breaking algorithm often succeeds in finding a way to
break a paragraph without needing hyphens at all, and since there's a good
chance of finding a different hyphen point near to one that is missed by
\TeX's patterns, it is clear that manual intervention to correct or insert
hyphenations in \TeX\ output is rarely needed, and that such refinements
take a negligible amount of time compared to the normal work of
keyboarding and proofreading.

But you can always insert words into \TeX's ^{exception dictionary}, if
you find that the patterns aren't quite right for your application.
For example, this book was typeset with three exceptional words added:
The format in Appendix~E includes the command
\begintt
\hyphenation{man-u-script man-u-scripts ap-pen-dix}
\endtt
which tells \TeX\ how to hyphenate the words `manuscript', `manuscripts',
and `appendix'. Notice that both singular and plural forms of
`manuscript' were entered, since the exception dictionary affects
hyphenation only when a word agrees completely with an exceptional
entry. \ (Precise rules for the ^|\hyphenation| command are
discussed below.)

If you want to see all of the hyphens that plain \TeX\ will find in some
random text, you can say `^|\showhyphens||{|\<random text>|}|' and the
results will appear on your terminal (and in the log file). For example,
\begintt
*\showhyphens{random manuscript manuscripts appendix}
|smallskip
Underfull \hbox (badness 10000) detected at line 0
[] \tenrm ran-dom manuscript manuscripts ap-pendix
\endtt
shows the hyphen positions that would have been found in this book
without the addition of any ^|\hyphenation| exceptions. Somehow the word
`manuscript' slips through all of the ordinary patterns; the author added it
as an exception for this particular job because he used it 80~times (not
counting its appearances in this appendix).

The |\showhyphens| macro creates an hbox that is intentionally
^{underfull}, so you should ignore the warning about `|badness 10000|';
this spurious message comes out because \TeX\ displays hyphens in compact
form only when it is displaying the contents of anomalous
hboxes. \ (\TeX\ wizards may enjoy studying the way |\showhyphens|
is defined in Appendix~B.)

\danger If you want to add one or more words to the exception dictionary,
just say ^|\hyphenation||{|\<words>|}| where \<words> consists of one
or more \<word> items separated by spaces. A \<word> must consist entirely
of letters and hyphens; more precisely, a ``hyphen'' in this context is
the token |-|\\{12}. A ``letter'' in this context is a character token whose
category code is 11 or~12, or a control sequence defined by ^|\chardef|,
or ^|\char|\<8-bit number>, such that the corresponding character has a nonzero
^|\lccode|. \TeX\ uses the |\lccode| to convert each letter to ``^{lowercase}''
form; a word-to-be-hyphenated will match an entry in the exception
dictionary if and only if both words have the same lowercase form after
conversion to lowercase.

\danger \TeX\ will henceforth insert discretionary hyphens in the specified
positions, whenever it attempts to hyphenate a word that matches an entry
in the exception dictionary, except that plain \TeX\ blocks hyphens after the
very first letter or before the last or second-last letter of a word.
You must insert your own discretionary hyphens if you want to allow them
in such positions. A |\hyphenation| entry might contain no hyphens at all;
then \TeX\ will insert no hyphens in the word.

\danger The exception dictionary is global; i.e., exceptions do not disappear
at the end of a ^{group}. If you specify |\hyphenation| of the same word more
than once, its most recently specified hyphen positions are used.

\ddanger The exception dictionary is dynamic, but the pattern dictionary
is static: To change \TeX's current set of hyphenation patterns, you
must give an entirely new set, and \TeX\ will spend a little time putting them
into a form that makes the hyphenation algorithm efficient. The
command format is ^|\patterns||{|\<patterns>|}|, where \<patterns> is a
sequence of \<pattern> items separated by spaces. This command
is available only in ^|INITEX|, not in production versions of \TeX, since
the process of pattern compression requires extra memory that can be put
to better use in a production system. |INITEX| massages the patterns and
outputs a format file that production versions can load at high speed.

\ddanger A \<pattern> in the |\patterns| list has a more restricted form
than a \<word> in the |\hyphenation| list, since patterns are supposed to
be prepared by experts who are paid well for their expertise. Each
\<pattern> consists of one or more occurrences of \<value>\<letter>,
followed by \<value>. Here \<value> is either a digit (|0|\\{12} to |9|\\{12})
or empty; an empty \<value> stands for zero. For example, the pattern
`{\tt\\0a\\1b\\0}' can be represented as |0a1b0| or |a1b0| or |0a1b| or
simply |a1b|. A \<letter> is a character token of category 11 or~12 whose
|\lccode| is nonzero. If you want to use a ^{digit} as a \<letter>, you
must precede it by a nonempty \<value>; for example, if for some reason you
want the pattern `{\tt\\1a\\01\\2}' you can obtain it by typing
`|1a012|', assuming that |\lccode`1| is nonzero. Exception: The character
`|.|'~is treated as if it were a \<letter> of code~0
when it appears in a pattern. Code~0 (which obviously cannot match a nonzero
|\lccode|) is used by \TeX\ to represent the
left or right edge of a word when it is being hyphenated.

\ddanger Plain \TeX\ inputs a file called |hyphen.tex| that sets up the
pattern dictionary and the initial exception dictionary. The file has the
form
\begindisplay
|\patterns{.ach4 .ad4der .af1t .al3t|\quad$\cdots$\quad|zte4 4z1z2 z4zy}|\cr
|\hyphenation{as-so-ciate as-so-ciates dec-li-na-tion oblig-a-tory|\cr
|  phil-an-thropic present presents project projects reci-procity|\cr
|  re-cog-ni-zance ref-or-ma-tion ret-ri-bu-tion ta-ble}|\cr
\enddisplay
The first thirteen exceptions keep \TeX\ from inserting incorrect
hyphens; for example, `|pro-ject|' and `|pre-sent|' are words like
`|re-cord|', that cannot be hyphenated without knowing the context.
The other exception, `|ta-ble|', is included just to meet the claim
that plain \TeX\ fully hyphenates the 700 or so most common words
of English.

\looseness-1
\ddanger But how does \TeX\ decide what sequences of letters are ``words''
that should be hyphenated? Let's recall that \TeX\ is working on a
horizontal list that contains boxes, glue, rules, ^{ligatures}, ^{kerns},
discretionaries, marks, whatsits, etc., in addition to simple characters;
somehow it has to pick out things to hyphenate when it is unable to
find suitable breakpoints without hyphenation. The presence of
punctuation marks before and/or after a word should not make a word
unrecognizable or unhyphenatable; neither should the presence of
ligatures and kerns within a word. On the other hand, it is desirable to
do hyphenation quickly, not spending too much time trying to handle
unusual situations that might be hyphenatable but hard to recognize
\hbox{mechanically}.

\ddanger \TeX\ looks for potentially hyphenatable words by searching
ahead from each glue item that is not in a math formula. The search
bypasses characters whose |\lccode| is zero, or ligatures that begin
with such characters; it also bypasses whatsits and {\sl^{implicit kern}\/}
items, i.e., kerns that were inserted by \TeX\ itself because of information
stored with the font. If the search
finds a character with nonzero |\lccode|, or if it finds a ligature
that begins with such a character, that character is called the
{\sl starting letter}. But if any other type of item occurs before a suitable
starting letter is found, hyphenation is abandoned (until after the next
glue item). Thus, a box or rule or mark, or a kern that was explicitly
inserted by ^|\kern| or~|\/|, must not intervene between glue and a
hyphenatable word. If the starting letter is not lowercase (i.e.,
if it doesn't equal its own |\lccode|), hyphenation is abandoned unless
^|\uchyph| is positive.

\ddanger If a suitable starting letter is found, let it be in font~$f$.
Hyphenation is abandoned unless the ^|\hyphenchar| of~$f$ is a number
between 0 and~255, inclusive.
If this test is passed, \TeX\ continues to scan forward
until coming to something that's not one of the following three
``admissible items'': (1)~a character in font~$f$ whose |\lccode|
is nonzero; (2)~a ligature formed entirely from characters of type~(1);
(3)~an implicit kern. The first inadmissible item terminates this part of
the process; the trial word consists of all the letters found in admissible
items, up to a maximum of~63. Notice that all of these letters are in font~$f$.

\ddanger If a trial word $l_1\ldots l_n$ has been found by this process,
hyphenation will still be abandoned unless $n\ge\lambda+\rho$, where
$\lambda=\max(1,\hbox{|\lefthyphenmin|})$ and
$\rho=\max(1,\hbox{|\righthyphenmin|})$.
^^|\lefthyphenmin| ^^|\righthyphenmin|
\ (Plain \TeX\ takes $\lambda=2$ and $\rho=3$.) \ Furthermore, the items
immediately following the trial word must consist of zero or more
characters, ligatures, and implicit kerns, followed immediately by
either glue or an explicit kern or a penalty item or a whatsit or an
item of vertical mode material from ^|\mark|, ^|\insert|, or ^|\vadjust|.
Thus, a box or rule or math formula or discretionary following too closely
upon the trial word will inhibit hyphenation. (Since \TeX\ inserts
empty discretionaries after ^{explicit hyphens}, these rules imply that
already-hyphenated compound words will not be further hyphenated by
the algorithm.)

\ddanger Trial words $l_1\ldots l_n$ that pass all these tests are submitted
to the hyphenation algorithm described earlier. Hyphens are not inserted
before $l_\lambda$ or after $l_{n+1-\rho}$.
If other hyphenation points are found,
one or more discretionary items are inserted in the word; ligatures
and implicit kerns are reconstituted at the same time.

\ddanger Since ligatures and kerns are treated in quite a general manner,
it's possible that one hyphenation point might preclude another because the
ligatures that occur with hyphenation might overlap the ligatures
that occur without hyphenation. This anomaly probably won't occur in real-life
situations; therefore \TeX's interesting approach to the problem will
not be discussed here.

\ddanger According to the rules above, there's an important distinction
between implicit and explicit kerns, because \TeX\ recomputes implicit
kerns when it finds at least one hyphen point in a word. You can
see the difference between these two types of kerns when \TeX\ displays
lists of items in its ^{internal format}, if you look closely:
`|\kern2.0|' denotes an implicit kern of $2\pt$, and `|\kern 2.0|'
denotes an explicit kern of the same magnitude. The ^{italic correction}
command~|\/| inserts an explicit kern.

\ddanger The control sequence ^|\-| is equivalent to
|\discretionary{\char|$\,h$|}{}{}|, where $h$ is the
^|\hyphenchar| of the current font, provided that $h$ lies
between 0 and~255. Otherwise |\-| is equivalent to
|\discretionary{}{}{}|.

\ddanger \looseness=-1
So far we have assumed that \TeX\ knows only one style of
hyphenation at a time; but in fact \TeX\ can remember up to 256
distinct sets of rules, if you have enough memory in your computer. An
integer parameter called ^|\language| selects the rules actually used;
every ^|\hyphenation| and ^|\patterns| specification appends new rules
to those previously given for the current value of\/ |\language|.
\ (If\/ |\language| is negative or greater than 255, \TeX\ acts as if
|\language|$\null=0$.) \ All |\patterns| for all languages must be
given before a paragraph is typeset, if |INITEX| is used for
typesetting.

\ddanger \TeX\ is able to work with several languages in the same
paragraph, because it operates as follows. At the beginning of a
paragraph the ``current language'' is set equal to |\language|. Whenever a
character is added to the current paragraph (i.e., in unrestricted
horizontal mode), the current language is compared to |\language|; if
they differ, the current language is reset and a whatsit node
specifying the new current language is inserted before the character.
Thus, if you say `|\def\french{\language1...}|' and `|mix| |{\french
franc/ais}| |with| |English|', \TeX\ will put whatsits before the |f|
and the~|w|; hence it will use language~1 rules when hyphenating
|franc/ais|, after which it will revert to language~0. You can insert
the whatsit yourself (even in restricted horizontal mode) by saying
^|\setlanguage|\<number>; this changes the current language but it
does not change |\language|. Each whatsit records the current
|\lefthyphenmin| and |\righthyphenmin|.
^^{French}

\endgroup
\endchapter

If all problems of hyphenation have not been solved,
at least some progress has been made since that night,
when according to legend, an RCA Marketing Manager received
a phone call from a disturbed customer. His 301 had just hyphenated ``God.''
\author PAUL E. ^{JUSTUS}, {\sl There's More to Typesetting Than %
  Setting Type\/} (1972)
% in IEEE Transactions on Professional Commun. vol PC-15, pp. 13-15

\bigskip

The committee skeptically re-
commended more study for a bill
to require warning labels on rec-
ords with subliminal messages re-
corded backward.
\author THE PENINSULA ^{TIMES TRIBUNE} (April 28, 1982)

\eject
\beginchapter Appendix I. Index

\maxdepth=4pt
The author has tried to provide as complete an index as possible, so that
people will be able to find things that are tucked away in obscure
corners of this long book. Therefore the index itself is rather long. A
short summary of the simpler aspects of \TeX\ appears at the beginning of
Appendix~B; a summary of special symbols appears at the end of Appendix~F;
a summary of other special things appears under `tables' below.

\medskip\ninepoint
Page numbers are \underbar{underlined} in the index when they represent
the definition or the main source of information about whatever is being
indexed. \ (Underlined entries are the most definitive, but not
necessarily the easiest for a beginner to understand.) \ A page number is
given in italics (e.g., `{\it123\/}') when that page contains an instructive
example of how the concept in question might be used.  Sometimes both
underlining and italics are appropriate.  When an index entry refers to a
page containing a relevant exercise, the answer to that exercise (in
Appendix~A) might divulge further information; an answer page is not
indexed here unless it refers to a topic that isn't included in the
statement of the relevant exercise.

\smallskip
Control sequence names that are preceded by an asterisk (*) in this index are
primitives of \TeX; i.e., they are built in. It may be dangerous to
redefine them.

\begindoublecolumns
\eightpoint \baselineskip=9.8pt minus .2pt % shooting for 53 or 54 lines/page
\parskip=0pt plus .8pt
\newdimen\sqht \sqht=2.4pt % \square
\newbox\astbox \setbox\astbox=\hbox to0pt{\hss\lower1pt\hbox{*}}
\raggedright \tolerance=5000 \hbadness=5000 \parfillskip 0pt plus 3em
\ttglue=.4em
\def\<#1>{\leavevmode\hbox{$\mkern-2mu\langle$#1\/$\rangle$}}
\let\oldcstok=\cstok
\def\cstok{\leavevmode\kern-2pt\oldcstok}
\def\vdots{\vbox{\baselineskip 3\p@\kern2\p@\hbox{.}\hbox{.}\hbox{.}}}
\def\ddots{\mathinner{\mskip1mu\raise5\p@\vbox{\kern2\p@\hbox{.}}\mskip2mu
    \raise3\p@\hbox{.}\mskip2mu\raise\p@\hbox{.}\mskip1mu}}
\def\cong{\mathrel{\mathpalette\@v@req\sim}} % congruence sign
\def\@v@req#1#2{\lower.5\p@\vbox{\baselineskip\z@skip\lineskip-.0\p@
    \ialign{$\m@th#1\hfil##\hfil$\crcr#2\crcr=\crcr}}}
\def\see{{\sl see\/}~\ignorespaces}
\def\also{\hfil\penalty50\hfilneg{\sl see~also\/}~\ignorespaces}
\let\oldttv=\ttverbatim
\let\+=\relax
\def\sub{\penalty100 \vskip -\parskip \quad}
\def\LaTeX{L\kern -.36em\raise.6ex\hbox{\sixrm A}\kern-.15em\TeX}
\def\ttverbatim{\oldttv \catcode`\*=\other \catcode`\,=\other
  \catcode`\.=\other \catcode`\;=\other \catcode`\@=\other \catcode`\+=\other}
\let\comma=, \let\period=. \let\asterisk=* \let\semicolon=;
\def\backup{\leavevmode\kern-1pt}\catcode`\-=11
\catcode`\*=\active \def*{\leavevmode\copy\astbox}
\catcode`\,=\active \def,{\eightrm\comma}
\catcode`\;=\active \def;{\/\eightrm\semicolon}
\catcode`\.=\active \def.{\eightrm\period\par\hangindent 2em }
\catcode`\+=\active \def+#1{\ifcat0\noexpand#1$\setbox\z@\hbox{#1}\dp\z@=\z@
   \underline{\box\z@}$\let\next=+%
  \else\let\next=#1\fi \next}
\catcode`\@=\active \let@=\eightit
\parindent=0pt
\hyphenpenalty=10000 \exhyphenpenalty=10000
\def\newletter{\medbreak\hangindent 2em}
\hangindent 2em
\] (visible space), +3, 420, 429; \also \<space>, spaces.
*|\|\] (control space), @8, 10, 19, @73, @74, 86--87, 154, @163, @167, 283,
  +285, +290, 323, 351, @381.
|#| (hash mark), 38, 51, 113, @200--202, @+203, @204--205, 228, +235, @236--240.
|\#| ( \#\ ), 38, 51, +356.
|##|, 203--205, 228, @359--362, @378--379.
|###|, 88.
|#{|, 204, @400.
|$| (dollar sign), 4, 38, 51, 54, 86--88, @92, 127, @134--135, @185--186,
  +269, 283, +287, +293.
|\$| ( \$\ ), 38, 51, @202, @309, +356.
|$$|, 86--89, +185, @186--196, @232, +287, +293, @375--376, @421.
|%| (percent sign), @26, 38, 39, 43, 48, 51, 113, 124, @249, @337, @340, +343.
|\%| ( \%\ ), 38, 43--44, 51, +356.
|%%|, 112--113.
|&| (ampersand), 38, 51, @175--177, @190--196, @231--248, +282, 385--386.
\sub for preloaded formats, +25, 26, 344.
|\&| ( \&\ ), 38, 51, 53, +356.
|&&|, 241--242, @361, @412.
|'| (apostrophe or right quote), 3--5, 51, @130, 155, 201, @305, 324,
  +357, 394--395; \also octal.
|\'| (acute accent), @7--9, 52--53, @305, @335, +356, @420.
|''| ( '' ), @3--5, @24, 394--395.
|`| (reverse apostrophe or left quote), 3--5, 51, 132, 134, @305, 391, 394--395;
  \also alphabetic constants.
|\`| (grave accent), 8, 52--53, @305, +356.
|``| ( `` ), @3--5, @24, 394--395.
|"| (double quote or ditto mark), 52, 53, 134; \also hexadecimal.
|\"| (dieresis or umlaut accent), @7, 9, @24, @25, 52--53, 55, +356.
|(| (left parenthesis), 51, 134, @140, @145--150, 345.
|\(|, 409.
|)| (right parenthesis), 51, 134, @140, @145--150, 345.
|[| (left bracket), 51, 134, @146--148, 171, @408, @437.
|[]|, +28, @+79, @302.
|[1]| (progress report), 23, +119.
|]| (right bracket), 51, 134, @146--147, 171, 345, @408, @437.
|{| (left brace), @13--14, @19--21, 38, 51, @200--202, +203--+204, @205--206,
  216, +269, 275--276, +283, +286, +291, @330.
|\{| ( $\{$ ), 134, @146--147, @174--175, +361.
|{}|, @19, @54, 82, 95, 114, @129, 130, 150, 169, @196, @242, 253, 262,
  @305, @315, @318, 351, 393.
|}| (right brace), @13--14, @19--21, 38, 51, @200--202, +203--+204, @205--206,
  +269, 275--276, +279, 301, @330.
|\}| ( $\}$ ), 134, @146--147, @174--175, +361.
\goodbreak
|+| (plus sign), 51, 132, 268.
|\+| (begin tabbed line), @231--234, 249, 339, +354.
|-| (hyphen or minus), @4, 51, 93, 95, @127, 132, 268.
*|\-| (discretionary hyphen), 95, 283, 287, 292, +455.
|--|, |---|, \see en-dash, em-dash.
$\pm$, \see |\pm|.
$\mp$, \see |\mp|.
|*| (asterisk), 23, 25, 51, 99, 113, 116, @132--133, 154, @326.
|\*| (discretionary $\times$), 173, +357.
|**|, 23, 25, 344.
|/| (slash), 51, @132, @146--147, @320, 430.
*|\/| (italic correction), @14, @64, +287, +292, @306, 382, 455.
\| (vertical line), 52, 53, 132, @146--147, @171, @174, 438.
|\|\| ( $\Vert$ ), @146--147, @171, +361, 435, 438.
|\| (backslash), @7, 38, 39, 40, 51, 146--147, +343, @436.
|\\|, 38, @356, @378, @418.
|<| (less than sign), 52, 53, @133, 150, 154, 209.
\leavevmode{\tt\rlap<\char`\_}, 45, 135, 368--369; \also |\le|.
|=| (equals sign), 51, @133, 209, 226, 275, @376.
|\=| (macron accent), 52, @53, +356.
$\buildrel \rm def \over=$, 181, @437.
\leavevmode{\tt\rlap/=}, 45, 135, 368--369; \also |\ne|.
|>| (greater than sign), 52, 53, @133, 150, 209.
|\>| (medium space), 167, 171, +357.
\leavevmode{\tt\rlap>\char`\_}, 45, 135, 368--369; \also |\ge|.
\<>, \see angle brackets.
|,| (comma), 51, @72--73, @134, @161--162, @172--174, @394--395, 430.
|\,| (thin space), 5, @167--173, 305, +357, @409--410.
|.|\ (period), 51, @72--73, 133--134, 149, @161, +345, @394--395, 430.
\sub space after, @73--75, +76.
|\.|\ (dot accent), 52, +356.
$\ldots\,$, \see |\ldots|, ellipses.
$\cdots\,$, \see |\cdots|, ellipses.
\leavevmode\smash{$\vdots\,$}, \see |\vdots|, ellipses.
|;| (semicolon), 51, @134, 161.
|\;| (thick space), 167, 171, +357.
|:|\ (colon), 51, @133--134, @155, 161, @174, @438.
$:=$, 133.
|?|\hbox{ }(question mark), 31, 51, 73, 161.
|?`|\hbox{ }(open question), 51.
|!|\hbox{ }(exclamation point), 51, @72, 73, 75, @169.
|!`|\hbox{ }(open exclamation), 51.
|\!|\ (negative thin space), 167, @169, +357.
|_| (underscore), 38, 51, @128--130, 134.
|\_| ( \_ ), 38, @165, +356.
|^| (hat), 38, 51, @128--130, 134, 369, @423.
|\^| (circumflex accent), 52--53, +356.
|^^|, 45, +47, @48, +368, 370, 423.
|^^M| (ASCII \<return>), 45, 249, 331, +343, 345, @348, @352, 380, @390--391,
  @421, 423.
|\^^M|, 8, @305, +351.
|~| (tilde), 38, 51, 343, +353; \also ties.
|\~| (tilde accent), 52, +356, 387.
{\tt\char'13} and {\tt\char'14}, 135, 343, 368--369, 429; \also |\uparrow|, |\downarrow|.
|@| (at sign), 51, 98--99, 132, 134, @344, @364, @408, 414.
|@@|, 98--99.
|\@ne|, +345.
\newletter
|\aa| ( \aa\ ), @52, +356.
|\AA| ( \AA\ ), @52, +356.
abbreviations, @73--74, @340; \also macros.
*|\above| (general fraction), @143, 152, +292, 444--445.
*|\abovedisplayshortskip|, +189, 274, @348, @415.
*|\abovedisplayskip|, +189, +190, 194, 274, 291, @348, @415.
*|\abovewithdelims|, 152, +292, 444--445.
absolute value, @146, @149, @171, @175.
Acc atom, 158, 289, 443.
*|\accent| (general accent), 9, 54, 86, 283, +286.
accents (\'{} \`{} \"{} etc\period), 7, @52--53, @339, +356, 357, 427--429.
\sub as ligatures, 46, 54.
\sub in math, @135--137, 141, 164--165, +359, 435, +443.
\sub on top of accents, 136.
\sub table, 52, 135, 339.
|\active| (category 13), @241, +343, @395, @421.
active character, 37, 40, 209, 241, 307, 377, 380--381, 394--395.
active math character, 155, +289.
active spaces, 381, 394, 421.
|\acute| (math accent: $\acute x$), 135, +359.
acute accent (\'{}), \see |\'|, |\acute|.
|\address|, @403--404, +407.
*|\adjdemerits|, +98, 273, @314, @348.
*|\advance|, @21, @118--119, @218, @256, +276, @355.
|\advancepageno|, 256, @257, +362, @416.
|\ae| ( \ae\ ), vii, 17, 45--46, @52--53, @239, +356.
|\AE| ( \AE\ ), @52--53, +356.
*|\afterassignment|, 215, +279, @352, @364, @376, @400.
*|\aftergroup|, 215, +279, @363, @374, @377, 379.
`ain, \see reverse apostrophe.
al-Khw\^arizm\^\i, abu `Abd All\^ah Mu\d{h}ammad ibn M\^us\^a, 53.
|\aleph| ( $\aleph$ ), 9, +358, 435.
Alice, 4, 387, 394.
alignment displays, 190, 193, +291.
\<alignment material>, 282, 285, 291.
alignments, 231--249, +282, 302--303, 385--386, 392; \also tabbing.
Alka-Seltzer, 404--405.
all caps, \see |\uppercase|.
Allen, Todd Andrew, 377.
allocation, 121--122, +346, +347.
|\allocationnumber|, 346.
|\allowbreak|, @174, +353, @396.
|\allowhyphens|, @394, +395.
|\alpha| ( $\alpha$ ), @127, @201, +358, 434.
|\Alpha|, 434.
alphabetic constants, @44--46, 215, +269, 270, 309, +367, 385.
alternatives, \see |\cases|.
|\amalg| ( $\amalg$ ), +358, 436.
American Mathematical Society, ii, vii.
ampersand, +25, 38, 51, @175--177, @190--196, @231--248, +282,
  @339, 344, 385--386, 428.
\AmSTeX, 164, +420, 434.
anatomy of \TeX, 38--39, +46, 85, +267, 349, 373, 379, 385, 386, +456.
|\angle| ( $\angle$ ), +358, 435.
angle brackets ( $\langle\,\rangle$ ), 59, @146--147, 150,~156, +268, 420, 437;
 \also \hbox to0pt{|\langle|, |\rangle|\period\hss}\par\hangindent2em
angstrom unit, \see |\AA|.
|\annotations|, 403, @404, +407.
answers to the exercises, 305--337.
Antisthenes of Athens, 239.
apostrophe, 3--5, 51, @130, 155, 201, 324; \also octal.
|\approx| ( $\approx$ ), 128, 436.
Arabic, 66.
|\arccos| ( $\arccos$ ), 162, +361.
Archytas of Taras, 239.
|\arcsin| ( $\arcsin$ ), 162, 361.
|\arctan| ( $\arctan$ ), 162, 361.
Arenski\u\i, Anton Stepanovich, 410.
|\arg| ( $\arg$ ), 162, 361.
arguments, 33, 200--205, @263, 268, @375--380.
Aristippus of Cyrene, 239.
Aristophanes, 239.
Aristotle, 35.
arithmetic, 117--119, \see |\advance|, |\multiply|, |\divide|.
\<arithmetic>, +276.
arrays, @176--178, \see matrices.
arrows, @146--147, @182, @226, +437.
|\arrowvert| ( $\big\arrowvert$ ), 150, +359.
|\Arrowvert| ( $\big\Arrowvert$ ), 150, +359.
{\sl\backup The Art of Computer Programming}, 259--260.
as is, \see |\obeylines|, |\obeyspaces|, verbatim.
ASCII, 3, 43--45, 49, 214, 343, +367, 371.
\<assignment>, +275.
assignments, 275--278, 373.
|\ast| ( $\ast$ ), 436.
asterisk, 23, 25, 51, 99, 113, 116, @132--133, 154, @326.
|\asymp| ( $\asymp$ ), 436.
|at|, 16--17, 60, 213, 277, @408, @414, 433.
\<at clause>, +277.
at sign, 51, 98--99, 132, 134, @344, @364, @408, 414.
AT\&T, 247.
atoms, 157--159, 170--171, 289--290, 441--447.
\sub table of atomic types, 158.
*|\atop|, @143, @145, 152, @178, +292, +444.
*|\atopwithdelims|, @152, +292, @324, @360, +444.
author, typesetting by, 182, 412--413.
auxiliary spaces, \see ties.
axis line, 150--152, 179, 443--447.
\newletter
|\b| (bar-under accent), 52, +356.
Bach, Johann Sebastian, 408.
Bach, P\period\ D\period\ Q\period, 410--411, 481.
backslash, @7, 38, 39, 40, 51, 146--147, 343, @436.
|\backslash| ( \backup$\backslash$ \backup), @38, @146--147, +359, 435, @436.
backspacing, @66, 82--83, 222, @394--395, @418.
Backus, John Warner, 268.
Bacon, Francis, viscount St\period\ Albans, 41.
Bacon, Leonard, 1.
bad breaks, avoiding, 27--30, 91--94, 173--174, 197.
badness, @28--30, 97--99, 111--113, 302.
*|\badness|, 214, @229, 271.
\<balanced text>, 275, +276, 385.
balancing columns, 386--388, @396--397, @417.
|\bar| (math accent: $\bar x$), 135, @136.
bar accent (\={}), \see |\=|, |\bar|.
bar-under accent (\b{\hbox{ }}), \see |\b|.
Barrett, Percy Reginald, 197.
Barrough, Philip, 229.
baseline, 15, +63, 77, 80--81, 150.
*|\baselineskip| (normal vertical distance between baselines), @78--79, +80,
  104, 194, 253, @256, 274, 281, 342, 349, @351--352, @409, @414--415.
*|\batchmode|, +32, +277, 299, @336.
Batey, Charles, 197.
beauty, 1.
Beck, Simone, 233, 236.
Beethoven, Ludwig van, 408, 410--411.
Beeton, Barbara Ann Neuhaus Friend Smith, 483.
|\beginchapter|, 418.
*|\begingroup|, 21, @249, @262, +279, @380, @407, @419.
|\beginsection|, @340--341, +355.
*|\belowdisplayshortskip|, +189, 274, @348, @415.
*|\belowdisplayskip|, +189, +190, 194, 274, 291, @348, @415.
Bemer, Robert William, \see TEX, ASCII.
bent bars, \see angle brackets.
Bertholle, Louisette, 233, 236.
|\beta| ( $\beta$ ), @127, 434.
|\Beta|, 434.
|\bf| (use {\bf boldface} type), @13--14, @164--165, @328, +351, @409,
  @414--415.
|\bffam|, +351, @414--415.
|\bgroup| (implicit |{|), 269, +351, @363, 382, +385, @407, @421.
Bibby, Duane Robert, i.
Biblical references, 303, @311.
bibliographies, 4, @74, 93, @340--341.
|\big| (largish delimiter), 147, 171, @320, @359, +360, @414--415.
|\Big| (between |\big| and |\bigg|), 147, 175, @359, +360.
big-$O$ notation, 132, @161--162, @169.
big point, 57, \see |bp|.
|\bigbreak|, 111, 116, +353, @363.
|\bigcap| (large $\cap$), 147, 435.
|\bigcirc| ( $\bigcirc$ ), 436.
|\bigcup| (large $\cup$), 147, 435.
|\bigg| (large delimiter), @147, 175, 196, 327, @359, +360.
|\Bigg| (larger than |\bigg|), 147, 175, @359, +360.
|\bigggl|, |\bigggr|, 324.
|\biggl| (|\bigg| left delimiter), 147, @149, +359, 437.
|\Biggl| (|\Bigg| left delimiter), 147, 149, +359, 437.
|\biggm| (|\bigg| middle delimiter), 147, +360.
|\Biggm| (|\Bigg| middle delimiter), 147, +360.
|\biggr| (|\bigg| right delimiter), 147, @149, +359.
|\Biggr| (|\Bigg| right delimiter), 147, 149, +359.
|\bigl| (|\big| left delimiter), @146--147, 149--150, 155, 171, @175, +359, 437.
|\Bigl| (|\Big| left delimiter), 147, +359, 437.
|\bigm| (|\big| middle delimiter), 147, 171, @175, +359.
|\Bigm| (|\Big| middle delimiter), 147, +359.
|\bigodot| (large $\odot$), 435.
|\bigoplus| (large $\oplus$), 435.
|\bigotimes| (large $\otimes$), 435.
|\bigr| (|\big| right delimiter), @146--147, 149--150, 171, @175, +359.
|\Bigr| (|\Big| right delimiter), 147, +359.
|\bigskip|, 70, 109, 111, 115--116, +352, @355, @407, @410--412.
|\bigskipamount|, 123, +349, @352--353, @363.
|\bigsqcup| (large $\sqcup$), 435.
|\bigtriangledown| ( $\bigtriangledown$ ), 436.
|\bigtriangleup| ( $\bigtriangleup$ ), 436.
|\bigtype|, 408--409, 411.
|\biguplus| (large $\uplus$), +358, 435.
|\bigvee| (large $\vee$), 435.
|\bigwedge| (large $\wedge$), 435.
Bill, 387, 394.
Bin atom, 158, 170--171, 289, 442--444, 446.
binary operations, @132--133, 154--155, 164, 196, 358, 435, +436;
  \also Bin atom.
binary search, 387--388.
binomial coefficient, 143, \see |\choose|.
*|\binoppenalty|, 101, 174, 272, 322, @348, +446.
black box, 64, +221, +222.
blackboard bold (e\period g\period, $\rm I\!R$), 164, 434.
blank line in input file, 24, 37, +47, @340--341, 381.
blank space, \see spaces.
Blase B\"ohning, Maria Dorothea, 248.
block structure, \see grouping.
block style, 405--407.
|\bmit| (boldface math italic), 156.
|\bmod| ( $\bmod$ ), @164, 322, +361.
|\body|, @403--404, +407.
Boehm, Peter James, 159.
B\"ohning, Jobst Heinrich, 248.
B\"ohning, Martin John Henry, 248.
Bohning Knuth, Louise Marie, 248.
boldface, 13, 156, 164--165, 386.
book design, 412.
book preparation, 303, 425.
|\bordermatrix|, @177, +361.
|\bot| ( $\bot$ ), 435.
*|\botmark|, 213, +258, 259--260, @262--263, 280.
boundary item, 157, 442, 446.
Bourbaki, Nicolas, 106.
|\bowtie| ( $\bowtie$ ), +358, 436.
\<box>, 120, 222, +278, 282, 285, 290.
*|\box| (use box register), 120--122, 151, 222, +278, 346, @354, @386, @387.
\sub |\box255|, 125, 253--258.
\<box dimension>, +271, 277.
box displays, 66, 75, 79, 158--159, 302, 455.
box memory, 300, 394.
\<box or rule>, +281.
\<box size assignment>, +277.
\<box specification>, 222, +278.
\cstok{boxed} control sequence names, 38.
boxed material, @223, @420.
boxes, 63--67, 77--83, 221--229.
|\boxit|, 223, 331.
*|\boxmaxdepth|, +81, 113, 249, @255, 274, @348.
|bp| (big point), +57, 270.
|\brace| (notation like $n\brace k$), +360.
|\braceld|, |\bracelu|, |\bracerd|, |\braceru| (pieces of horizontal braces),
  @+357.
braces, 51, 216, +269, 275--276, 279, 283, 286, 289--291, 330, 345, 385--386.
\sub for arguments to macros, @20, @200--202, +203, +204, @205--206, 385--386.
\sub for grouping, @13--14, @19--21, 232, 248, 253.
\sub horizontal, @176, 225--226, @339.
\sub implicit, +269, \see |\bgroup|, |\egroup|.
\sub in math formulas, @145--147, @174--176.
|\bracevert| ( $\big\bracevert$ ), 150, +359.
|\brack| (notation like $n\brack k$), +360.
brackets, 51, 134, @146--148, 171, @408, @437.
|\break| (force line or page break), 94, 97, 106, 114, 193, +353.
breakpoints, +96, 97--100, +110, 111--114, 394.
\sub avoiding bad, 27--30, 91--94, 109--111, 173--174, 197.
\sub discretionary, @95--96, 173, +287, +292, @357.
\sub forcing good, 94, 105, 109--111, 114.
\sub in displays, 195--197.
\sub in formulas, 173--174, 446--447.
|\breve| (math accent: $\breve x$), 135.
breve accent (\u{}), \see |\u|, |\breve|.
British pound sign, 54.
brochures, 251.
*|\brokenpenalty|, +104, 105, 272, @348.
Brooks, Frederick Phillips, Jr\period, 365.
Brown, Peter John, 425.
|\buildrel|, +361, @437.
built-up (extensible) characters, 442.
built-up fractions, \see |\over|.
|\bull| ( $\bull$ ), +420.
Bull, John, 239.
|\bullet| ( $\bullet$ ), @133, 154, @355, 436.
bulleted lists, 102, 105.
business correspondence, 200, 403--408.
|by|, @118, +276.
|\bye|, 87--88, @340, +357.
Byron, George Gordon No\"el Byron, baron, vii.
\newletter
|\c| (cedilla accent), @24--25, 52, +356.
|\cal| (calligraphic caps), 164, +351, 431, 434.
calculus, 168--169, 180--181.
camera alignment, 416--417.
|\cap| ( $\cap$ ), 133, 436.
capacity of \TeX, 100, 300--301, 383.
caps and small caps, 203.
captions, 115.
caret, 369.
caron, \see h\'a\v cek.
carriage return, 231, \see \<return>, |\cr|.
|\cases| ( $\bigl\{{\cdots\atop\cdots}$ ), 175, +362.
Caswell, Herbert Ernest, 413.
*|\catcode|, @39, 134, 214, +271, 305, @343, @380--382, @384, @390--391,
  @421, @424.
category codes, 37--40, +47, 48, 203--205, 209--210, 214, 381.
\sub table, 37.
|cc| (cicero), +57, 270.
|\cdot| ( $\cdot$ ), @133, @172, 319, 436.
|\cdotp|, 358, +359, 438.
|\cdots| ( $\cdots$ ), @172, @176, @180--181, +359, 438.
cedilla accent (\c{\ }), 25, 52, 54, \see |\c|.
ceiling brackets ( $\lceil\,\rceil$ ), 146--147, \see |\lceil|, |\rceil|.
centering, 71, 233, 236.
|\centering|, +347, 348, 362.
|\centerline| (make a centered line), @20, @24, 33, 71, 85, 101, @117, 232,
  311, @340, +353.
|\cents| ( \hbox{\rm\rlap/c} ), 140, +319.
Ces\`aro, Ernesto, 53.
Cha\u\i kovski\u\i, P\"etr Il'ich, 410--411.
*|\char|, @43--45, 76, 86, 155, 283, +286, 289, @340, @427, 452.
\<character>, +289.
character codes, 43--46, 367--370; \also category codes.
\<character token>, +270.
*|\chardef|, 44, 121, 155, 210, 214, 215, 271, +277, 336, @343, @345, @356, 452.
\<chardef token>, +271, 283, 286, 289.
Charles XII of Sweden, 92.
Chaundy, Theodore William, 197.
|\check| (math accent: $\check x$), 135.
check accent (\v{}), \see |\v|, |\check|.
chemical typesetting, 179.
Cherry, Lorinda Landgraf, 159.
|\chi| ( $\chi$ ), 1, 434.
Chicago, University of, Press, 125, 293, 418, 439.
Child, Julia, 233, 236.
Children's Television Workshop, ii.
choice, four-way, 157, +292, +442.
|\choose| (notation like \setbox0=\hbox{$n\choose k$}\dp0=0pt\box0),
  139, 143, 152, 178, +360.
Christie Mallowan, Dame Agatha Mary Clarissa (Miller), 249.
cicero, \see |cc|.
|\circ| ( $\circ$ ), @133, @323, 436.
circles, \see |\circ|, |\bigcirc|.
circular quotation, 101.
circumflex, 369, \see hat.
circumflex accent (\^{}), 52, +356, \see |\^|.
classes of math characters, table, 154.
*|\cleaders|, +224, 225--226, @357, @374.
|\cleartabs|, @234, +354.
Close atom, 158, 170--171, 289, 443--444, 446.
*|\closein|, +217, 280.
*|\closeout|, 226--228, 254, +280, @422.
|\closing|, @403--404, +407.
closings, 134, 147, 154--155, 359, +437; \also Close atom.
club lines, 104, 272, 398.
*|\clubpenalty|, +104, 113, 272, 317, @348, @419.
|\clubsuit| ( $\clubsuit$ ), 435.
|cm| (centimeter), @24, +57, 270.
|cmbx| fonts, 60, 350, 413, 428, 433.
|cmex| fonts, 157, 225, 350, 432--433.
|cmmi| fonts, 350--351, 413--414, 430, 433.
|cmr| fonts, 16--17, 60, 63--64, 76, 350, 413, 427, 433.
|cmsl| fonts, 63--64, 350, 413, 428, 433.
|cmsy| fonts, 157, 350--351, 413--414, 431.
|cmti| fonts, 350, 413, 428, 433.
|cmtt| fonts, 60, 350, 413--414, 429, 433.
\<code assignment>, +277.
\<codename>, +271.
codes for characters, 43--46, 367--370.
Coelho, Manuel Rodrigues, 239.
coerce \<dimen> to \<number>, 270.
coerce \<glue> to \<dimen>, 270.
coerce \<number> to \<dimen>, 336.
\<coerced dimen>, +270.
\<coerced integer>, +269.
\<coerced mudimen>, +270.
collective signs, \see large operators.
colon, 51, @133--134, @155, 161, @174, @438.
|\colon| ( $\colon$ ), @134, +359, @438.
color, 229.
column vector, 177.
column width, 29, 231, @257, @387, 417.
|\columns|, 231, 354.
comma, 51, @72--73, @134, @161--162, @172--174, @394--395, 430.
commands, 267--293.
comments, 26, 47, 337, @340.
communication between macros, 211, 328--329, 375--376, 407--408.
commutative diagram, @182.
composing stick, 64--65.
compound fraction, @143.
compound matrix, @181.
Computer Modern fonts, 16, 350, 427--438.
computer programs, 38, 165, @234.
concert programs, 408--412.
conditionals, 206--208, +209, +210, 211, 240, @308, @384.
|\cong| ( $\cong$ ), 151, +360, 436.
constants, +269, 270, 308.
continued fractions, @142.
\<control sequence>, +275, 277.
control sequences, 7--11, +46, 199, 457.
\sub misspelled, 31--32.
control space (|\|\]), @8, 10, 19, @73, @74, 86--87, 154, @163, @167, 283,
  +285, +290, 351, @381.
control symbols, 7--8, 46--47.
control words, 7--8, 38, 46--47, 204.
coordinates, 389.
|\coprod| (large $\amalg$), 435.
*|\copy| (copy a box), 120, 151, +222, 278, @329, @374, @386, @407.
|\copyright| ( \copyright\ ), ii, @308, 339, +356.
Cornet, Peeter, 239.
Correa de Arauxo, Francisco, 239.
correspondence, 200, 403--408.
|\cos| ( $\cos$ ), 162, 361.
|\cosh| ( $\cosh$ ), 162, 361.
cost of a page break, 111--113, 124.
|\cot| ( $\cot$ ), 162, 361.
|\coth| ( $\coth$ ), 162, 361.
*|\count| registers, 118--122, 207--208, +271, +276, @346--347, @379.
\sub |\count0|, +119, @207, 252--254, +362.
*|\countdef|, 119, 121, 210, 215, 271, +277, @346--347.
\<countdef token>, +271.
Cowper, William, 35.
*|\cr| (end of aligned row), @175--177, @190--197, @231--238, +245, +248,
   275, 282, 351, 352, 385--386, @412, @418, @421.
\sub avoiding, 249.
cramped styles, 140--141, 445.
*|\crcr| (force |\cr|), 249, 275, 282, @361--362, 385, @412, @421.
Crocker, Betty, 425.
cross, \see |\dag| (~\dag~), |\times| (~$\times$~).
crotchets, \see brackets.
|\csc| ( $\csc$ ), 162, 361.
*|\csname|, 40--41, +213, @348, @375.
|\csname\endcsname|, 46, 308.
cube root, @130--131.
Cummings, Edward Estlin, 49.
|\cup| ( $\cup$ ), 133, 436.
curly braces, \see braces.
current font, 13, 20, 154, 163, 213--214.
current page, 112, 122--125, 278, 280.
cyclic preambles, +241, @242, @246, @361, @412.
Cyrillic characters, 370.
\newletter
|\d| (dot-under accent), 52--53, +356.
|\dag| ( \dag\ ), 53, 117, +356, 438--439.
|\dagger| ($\dagger$ as binary operator), 436, 439.
Dale, Robert William, 283.
|\danger|, 419.
dangerous bend, v--vi, 5, 15, 44--45, 70, 419.
Danish characters, 45--46, 52--53, 370.
|\dash|, 30.
dashes, 4, @26, @30, 51, 93, 95.
|\dashv| ( $\dashv$ ), 436.
date, today's, 406.
*|\day|, 273, 349, @406.
|dd| (didot point), +57, 270, 272.
|\ddag| ( \ddag\ ), 117, +356, 438--439.
|\ddagger| ($\ddagger$ as binary operator), 436, 439.
|\ddot| (math accent: $\ddot x$), 135, +359.
|\ddots| ( \smash{$\ddots$} ), 177, +359.
|\ddt| (debugging aid), 248.
De Vinne, Theodore Low, 107.
*|\deadcycles|, 214, +255, 264, 271, 283, @400.
debugging, 205, 248, 298--303, 347.
decent lines, 97, 99.
\<decimal constant>, +270.
decimal points, 57, 134, 240.
\<def>, +275.
*|\def|, 44, 136, 199--208, 215, 275--276.
default output routine, 253--255.
default rule thickness, 443--447.
default values of parameters, 348--349.
*|\defaulthyphenchar|, 273, @348.
*|\defaultskewchar|, 273, @348.
defining a control sequence, 199--208.
\<definition>, +275.
\<definition text>, +275.
|\deg| ( $\deg$ ), 162, 361.
degrees ( $^\circ$ ), @180.
*|\delcode|, +156, 214, 271, +290, @345.
\<delete> (ASCII code 127), 37, 39, 343, 369.
deleting tokens, 32, 215, @295--297.
\<delim>, 289--290.
delimited arguments and parameters, 203--205, 249, @263, @375--377, @407.
*|\delimiter|, +156, 289--290, @359.
*|\delimiterfactor|, 152, 273, @348, +446.
delimiters, 145--150, 156--157, @171, +290, 437, +442.
*|\delimitershortfall|, 152, 274, @348, +446.
|\delta| ( $\delta$ ), @127, 434.
|\Delta| ( $\Delta$ ), @169, @186, 434.
demerits, 94, 97--99, 273, @451.
denominator, 141, 152, 179, 444--445.
|depth|, 221, 224, +282, 337.
depth of a box, 63--67, 77, 80--82, 225.
Derek, Bo, 293.
descenders, 63, 113, 319, 323.
design size, 16--17, 213.
|\det| ( $\det$ ), 162, +361.
device-independent output, 23; \also |.dvi|.
Diabelli, Antonio, 408.
diagnostic form of lists, 66, 75, 79, 158--159, 302, 455.
dialogs with the user, 217--218.
|\diamond| ( $\diamond$ ), 436.
diamond leaders, viii.
|\diamondsuit| ( $\diamondsuit$ ), 435.
D\'\i az de la Pe\~na, Maximiliano Antonio Tem\'\i stocles, 384.
Dick and Jane, 72--74, 76.
Dictionaries, 259, 449--453.
Didot, Fran\c cois Ambroise, 57.
didot point, 57, \see |dd|.
dieresis (\"{}), +52, 53, 356, \see |\"|.
diesis ( \ddag\ ), \see |\ddag|.
Dieter, Ulrich Otto, 14.
\<digit>, +269.
\<digit string>, 57.
digits, 51, 132, 435, 453.
\sub width of, 60, 241, 427.
|\dim| ( $\dim$ ), 162, 361.
\<dimen>, 59, 61, 71, 118, +270, 271.
\<dimen parameter>, 271, +274, 276.
*|\dimen| registers, 118--122, +271, +276, @346--347, @349, @360, @363, @395.
\<dimen variable>, +276.
*|\dimendef|, 119, 215, +277, @346--347.
\<dimendef token>, +271.
dimensions, 57--61.
\sub as arguments, 204, 362.
Dionysius I of Syracuse, 239.
diphthongs, \see |\ae|, |\oe|.
direct sum, \see |\oplus|.
\<disc text>, +287, 292.
discardable items, 95, 110--112, 124, @393.
*|\discretionary|, 95--96, 283, 286, +287, +292.
discretionary hyphens, 28, 95--96, 453, +455.
discretionary multiplication signs, 173, +357.
display math mode, 85--89, 289--293.
display style, 140--142, 441--447.
*|\displayindent|, 188, 190, 274, 291, 349.
*|\displaylimits|, 144, 159, +292, 443.
|\displaylines|, 194, @196, +362.
displays, 87, 103, @139--145, @166--167, 185--197, @232, @241, 315.
\sub at beginning of paragraph, 316.
\sub non-centered, 186, 326, 375--376, 420--421.
\sub positioning of, +188--+190.
*|\displaystyle|, 141--142, +292, @362.
*|\displaywidowpenalty|, +104, 105, 272, @348.
*|\displaywidth|, 188, 190, 274, 349.
Disraeli [Beaconsfield], Benjamin, earl, 219.
ditto mark, 53, 441.
|\div| ( $\div$ ), 436.
*|\divide|, @118--119, @218--219, +276, @391, @397, @398, @417.
|\do|, 344, 380, 423.
dollar sign, 4, 38, 51, 54, 86--88, @92, 127, @134--135, @185--186, 283,
  +287, +293.
Donnelley, Richard Robert, vii.
|\dospecials|, +344, 380, 422--423.
|\dosupereject|, 256, +364.
|\dot| (math accent: $\dot x$), 135.
dot accent (\.{}), \see |\.|, |\dot|.
dot-under accent (\d{\hbox{ }}), \see |\d|.
|\doteq| ( $\doteq$ ), +361, 436.
|\dotfill| (\hbox to 4em{\dotfill}), 244, @334, @335, @340--341, +357.
dotless letters, 52--53, 136, 435.
|\dots| ( \dots\ ), 173, +356; \also ellipses.
double-column format, @257, 386--388, @417.
double dagger, \see |\ddag|, |\ddagger|.
double dangerous-bend signs, vi, 419.
double integrals, @169, @180.
double quote mark, 52, 53, 134; \also hexadecimal.
*|\doublehyphendemerits|, 98, 273, @348, @451.
|\downarrow| ( $\downarrow$ ), @146--147, @182, +359, 437.
|\Downarrow| ( $\Downarrow$ ), @146--147, +359, 437.
|\downbracefill| (\hbox to 4em{\downbracefill}), 225--226, +357.
Doyle, Sir Arthur Conan, 401.
*|\dp|, 120, 271, @316, +388--+389, @417.
dragon curve, 391.
Drofnats, Revinu Jitis, 24, 27--28, 73--74, 404--406, 408, 410--412.
Dudeney, Henry Ernest, 249, 334.
Dull, Brutus Cyclops, 131, 173, 260.
|\dummy|, 400.
*|\dump|, +283, +286, 336, @344.
Durant, William James, 239.
|.dvi|, 23, 43, 60, 119, 213, 228, 254, 279, 280, 302, 367, 374.
Dvo\v r\'ak, Anton\'\i n Leopold, 409.
$dx$, @168.
\newletter
EBCDIC, 43.
*|\edef|, @215--216, 275, @328, @373--374.
editing, 34, 139, 197.
efficiency, 329, 333, 342, 345, 347, 383, 384, 394, 399, 423.
|\egroup| (implicit |}|), 269, +351, @363, 382, +385, @407, @421.
Ehlert, Ernst Fred, 248.
Ehlert Bohning, Pauline Anna Marie, 248.
|\eightpoint|, +415, @416.
|\eject| (force page break), @24--25, 105, 109, 189, +353, @418, @419.
elbows, \see angle brackets.
|\ell| ( $\ell$ ), 132, 435.
ellipses ( $\cdots$ ), 73, 172--174, 176--177, 180--182, 245.
Elphinstone, Mountstuart, 89.
*|\else|, 207, 210, +213.
|em|, 60, 154, 166, 214, +270, @352, @414, 433.
em-dash ( --- ), 4, 302.
em quad, \see |\quad|.
embellished letters, \see accents.
emergency stops, 299--300.
*|\emergencystretch|, +107, 274.
Emerson, Ralph Waldo, 41.
emphatics, \see dot-under.
\<empty>, +268.
|\empty|, @263, +351, @378.
empty discretionary, 95, 286.
empty group, 19, 253, 305.
empty line in input file, 24, 37, +47, @340--341, 381.
\sub at end of file, 217.
empty line in output, 114, 316.
empty page, 114.
|\emptyset| ( $\emptyset$ ), 128, 435.
en-dash ( -- ), 4, 187, 252, 314.
en quad, 71, \see |\enspace|.
*|\end|, 23, 26, @27, 87, +264, +283, 286, 299, @336, @403.
end of an input file, 206, 214, 217.
end of file line, \see \<return>, |\cr|.
end of file page, \see \<formfeed>.
end of paragraph, 286, \see |\par|.
|\endchapter|, 418--419.
*|\endcsname|, 40--41, +213, 283, @348, @375.
endgame, 87, 264, 283.
|\endgraf|, @262, 286, @331, +351, @407, @416, @419.
*|\endgroup|, 21, @249, @262, +279, @380, @407, @419.
*|\endinput|, 47, +214.
|\endinsert|, 115--116, +363.
|\endletter|, @403--404, +407.
|\endline|, +351.
*|\endlinechar|, +48, 273, @331, +348, @390--391.
|\endtemplate|, 206, +240.
|\enskip|, 71, +352.
|\enspace|, @202, +352, @419.
enumerated cases, in formulas, 175.
\sub in separate paragraphs, @102, @340.
\sub within a paragraph, 92.
enunciations, \see |\proclaim|.
epigraphs, 418--419, 425.
|\epsilon| ( $\epsilon$ ), 1, 128, 434.
|\eqalign|, @190--191, 193, 242, 326, +362.
|\eqalignno|, @192--193, 194, +362.
\<eqno>, +293.
*|\eqno|, @186--187, 189--191, 193, +293, 375--376.
\<equals>, +275, 276--277.
equals sign, 51, @133, 209, 226, 275, @376.
equation numbers, 186--196.
|\equiv| ( $\equiv$ ), @133, 436.
Erd\H os, P\'al (= Paul), 53.
*|\errhelp|, 275, +280, @347.
*|\errmessage|, 216, +279--+280, @347, @418.
error messages, 30--33, 295--301, 308--309.
error recovery, 31--32, 46, 215, 295--303, 309.
*|\errorcontextlines|, +34, 273, @348.
*|\errorstopmode|, +32, +33, +277, 299.
es-zet ( \ss\ ), \see |\ss|.
\<escape> (ASCII code 27), 369.
escape character, 7, 37, 421.
*|\escapechar|, 40, 213, +228, 273, @308, +348, @377.
|\eta| ( $\eta$ ), 434.
|etc.|, 302.
Eulerian numbers, 152.
Evagoras of Salamis, 239.
even-numbered pages, 252--253, @416.
*|\everycr|, 275, @362.
*|\everydisplay|, 179, 275, +287, @326.
*|\everyhbox|, 275, 279.
*|\everyjob|, 275.
*|\everymath|, 179, 275, +287, +293, 326.
*|\everypar|, @105, 215, 253, @262, 275, +282, +283, @333, @381, @407, @421.
*|\everyvbox|, 275, 279.
|ex|, 60, 154, +270, @356, 433.
exception dictionary, 449, 452--453.
exclamation point, 51, @72, 73, 75, 161, @169.
|\exercise|, 10, +422.
exercises, vii, 1--303.
*|\exhyphenpenalty|, +96, @262, 272, @348.
|\exists| ( $\exists$ ), 435.
|\exp| ( $\exp$ ), 162, 361.
*|\expandafter|, @40, +213, 215, @260, @308, @330, @348, @374--380.
expansion of expandable tokens, 212--216, 238, 267, 373--374.
\sub avoiding, 216, 262--263, 377.
explicit hyphens, 4, 93, 454.
explicit kerns, 40, 280, 306, 454--455.
exponents, \see superscripts.
extensible characters, 442.
extension fonts, 157, @351, 432--433, 441, 447.
extensions to \TeX, 226, 228--229.
eyestrain, reducing, 59.
\newletter
faces, 13, 17, 390.
\<factor>, 270.
factorial, @169, @181.
*|\fam|, 154--159, 273, +289--+290, 346--347, @351, 358, @414--415.
families, 153--159, 289--290, 346, 431, 442.
\<family assignment>, +277.
\<family member>, +271, 277.
family tree, 248.
family 0 (math roman fonts), 153--157, @351.
family 1 (math italic fonts), 155--157, @351.
family 2 (math symbol fonts), 157, @351, 431, 441, 447.
family 3 (math extension fonts), 157, @351, 432--433, 441, 447.
{\tt Fatal format file error}, 299.
feasible breakpoints, 99, 315.
fences, \see openings, closings, delimiters.
ffl, \see ligatures.
*|\fi|, 207, 210, +213.
Fibonacci, Leonardo, of Pisa, 166.
fields of atoms, 158--159, 289--291.
|fil|, 72, 118--119, +271, @348, @394.
\<fil dimen>, +271.
\<fil unit>, +271.
|\filbreak|, 111, +353.
file names, 25, 214, 216--217, 226, 278.
file pages, 343.
file types, \see |.dvi|, |.fmt|, |.tfm|, |.tex|, log file, terminal.
\<filename>, 214, 216, 226, 277, +278.
|fill|, 72, 118--119, +271, @347.
fill page with blank space, \see |\vfill|.
\<filler>, +276, 278, 280--282, 289.
|filll|, 72, 118, +271, @332.
*|\finalhyphendemerits|, 98, @106, 273, @348, @451.
*|\firstmark|, 213, +258, 259--260, 280.
|\fiverm|, 153, +350, @351, @414--415.
|\flat| ( $\flat$ ), @409, 435.
floating insertions, 115--116, 125, @363.
*|\floatingpenalty|, +123--+124, 272, 281, @363.
floor brackets ( $\lfloor\,\rfloor$ ), 146--147, \see |\lfloor|, |\rfloor|.
flush left, 72, 142, 177, 181, 196.
flush right, 71--72, 106, 142, 177, 196, 233, 419.
|.fmt|, 39, 344.
|\fmtname| and |\fmtversion|, 364.
|\folio| (typeset page number), 252--253, +362, @406, @416.
\<font>, +213, 214, 271, 277.
*|\font|, 16--17, 60, 210, +213, 214--215, 271, +276.
\<font assignment>, +277.
font metric files, 46, 433.
\<font range>, +271.
\<fontdef token>, +271.
*|\fontdimen|, 76, 157, @179, 214, +271, +277, @355--356, @375, @390, +433,
  441, 447.
*|\fontname|, 213, 214.
fonts, 13--17; \also Computer Modern fonts.
|\footins|, 256, +363, 396--399, @416.
|\footline|, 252, 256, @340--341, +362.
|\footnote|, 82, 116, 251, 256, 340, +363, 382, 416.
|\footnoterule|, 256, +364.
footnotes, 105, @116--117, 121, 125, @173, 416--417.
\sub short, 395--399.
|\forall| ( $\forall$ ), 435.
forbidden control sequence, 206.
Ford, Patrick Kildea, 293.
foreign languages, 45--46, 52--54, 370, 387, 449, 451.
form letters, 200, 207--209.
format file, 25--26, 39, 283, 344.
format-independent documents, 194, 203.
formats, 11, 39, 200, 403--425, 434.
\<formfeed> (ASCII code 12), 343, 369.
formulas, 127--199.
Foss, Sam Walter, 439.
fractions, 67, 139--143, 152, 170, 179, 186, 332, 444--445.
\sub huge, 196.
\sub slashed form, 67, 139--140, 233, 236.
Fraktur, 164, 434.
Franklin, Benjamin, 65.
French, 54, 340--341, 455.
|\frenchspacing|, 74, @340, +351, 381, 400.
Frescobaldi, Girolamo, 239.
front matter, 413.
|\frown| ( $\frown$ ), 436.
full stop, \see period.
|\fullhsize|, 257.
|\fullline|, 257.
funny space, 377.
*|\futurelet|, +207, 215, @262, 277, @363, @375--377, @423.
|\futurenonspacelet|, 376.
\newletter
Galilei, Galileo, 101.
galley file, 303, 425.
Gamble, James, 38.
|\gamma| ( $\gamma$ ), @127, 434.
|\Gamma| ( $\Gamma$ ), @127, @169, +358, 434.
|\gcd| ( $\gcd$ ), 162, @192, 361.
*|\gdef|, +206, 215, 275, @352, @407.
|\ge| ( $\ge$ ), 9, 45, @175, 318, +361, 438.
\<general text>, +276, 279, 280.
generalized fraction, 152, 157, +292, +444--+445.
generic coding, 194, 203.
generic matrix, 177, 245.
|\geq| ( $\geq$ ), 318, 436.
German, 52, 96, 451.
German black letters, \see Fraktur.
|\getfactor|, @356, +375, @398.
|\gets| ( $\gets$ ), +361, 438.
|\gg| ( $\gg$ ), 436.
Gibbon, Edward, 117.
*|\global|, @21, 119, 206, @218, 232, 256, +275, 301, 307, 320, 346.
\<global assignment>, 179, +277.
*|\globaldefs|, 238, 273, +275.
glue, 63, 69--83, 95, 110, 157, 222--225, 302, 412.
\sub above and below displays, 189--190, 194.
\sub at top of page, 113--114, 124, 256.
\sub between aligned columns, 237--239, 247, 392.
\sub between lines, \see interline glue.
\sub between paragraphs, 79, 104--105, 262, 282, 342, 406, 417.
\sub between words, 74--76, 356, 393--394, 433.
\<glue>, 71, 118, +271.
|\glue|, 75, 79, 302.
\<glue parameter>, 271, +274, 276.
|glue set|, 79, 302.
glue set order, 77, 79, 81, 97.
glue set ratio, 77, 79, 81, 97.
\<glue variable>, +276.
goal height, 112--114, 123--125.
Goethe, Johann Wolfgang von, 183.
|\goodbreak|, 111, 116, +353.
Grandmaster, 253.
|\grave| (math accent: $\grave x$), 135.
grave accent (\`{}), \see |\`|, |\grave|.
greater than or equal, \see |\ge|.
greater than sign, 52, 53, @133, 150, 209.
Greek, 127--128, 137, 156, 164, 319, 358, 430, 434.
Green, Walter, 244.
grouping, 13--14, 19--21, 119--120, 122, 200, 201, 241, 259, +279, 283,
  286, 291, 301, 375, 378, 385, 453.
\sub characters for, 39--40, @381--382.
\sub implicit, 115, 148, 194, 253, +287, +292, +293.
guide words, 259.
\newletter
|\H| (long Hungarian umlaut), 52, @53, +356, @420.
H\&J, \see hyphenation, line breaking, setting glue.
h\'a\v{c}ek accent (\v{}), \see |\v|, |\check|.
halftones, 228.
*|\halign|, 117, 190, 193, 194, 235--249, +282, 286, +291, 302, @326, 352,
  @361--362, @386, @392.
\sub compared to tabbing, 235.
Halmos, Paul Richard, 183.
Hamza, \see apostrophe.
|\hang|, +355, @419.
*|\hangafter|, +102, 103--104, 273, 348--349, @419.
*|\hangindent|, +102, @262, 274, 349, @407.
hanging indentation, 79, 102--103.
hanging punctuation, 394--395.
Hardy, Godfrey Harold, 321.
harpoons, 437.
hash mark, 38, 51, 113, @200--202, @+203, @204--205, 228, +235, @236--240.
Hassler, Hans Leo, 239.
hat, 38, 51, @128--130, 134, 369, @423.
|\hat| (math accent: $\hat x$), @135--136, 164.
hat accent (\^{}), \see |\^|, |\hat|, |\widehat|.
*|\hbadness|, 29, 272, +302, @348, @387--388, 400.
|\hbar| ( $\hbar$ ), @169, +358, 435.
hbox (box with horizontal list inside), 64.
*|\hbox|, 64--67, 77, 86, 93, 151, 159, 163, 175, 179, 185--186, 221, 222,
  278, 282, 388--389.
|\headline|, 252--253, 255, +362, @406.
|\heartsuit| ( $\heartsuit$ ), 435.
Hebrew, 66.
|height|, 221, 224, +282, 337.
height of a box, 63--67, 77, 80--82, 225.
help messages, 32, 280, 295--296.
\<hex digit>, +269.
\<hexadecimal constant>, +269.
hexadecimal notation, +43, @44, 45, 47--48, @154, 219, 336, 420.
*|\hfil|, 71--72, 194, 235--237, 283, 285, 290, 397.
*|\hfill|, 71--72, 142, 177, 194, 233, 283, 285, 290.
*|\hfilneg|, 72, 100, 283, 285, 290, 397.
*|\hfuzz|, 30, 274, +302, @348, @387--388.
|\hglue|, +352.
|\hideskip|, +347, 348, 354.
|\hidewidth|, 243, 245, 247, 325, +354.
higher-level languages for composition, 203.
Highton, Albert Henry, 481.
Hilbert, David, 167.
*|\hoffset|, +251, 274, @342.
*|\holdinginserts|, +125, 273, 400.
Holmes, Thomas Sherlock Scott, 401.
|\hom| ( $\hom$ ), 162, 361.
Hombert, Humbert, 401.
Honeywell Information Systems, 1.
|\hookleftarrow| ( $\hookleftarrow$ ), 358, 437.
|\hookrightarrow| ( $\hookrightarrow$ ), 358, 430, 437.
hooks, \see |\subset|, |\supset|.
horizontal braces, @176, 225--226, @339.
\<horizontal command>, +283.
horizontal lists, 64, +94--+95.
horizontal mode, 85--89, 105, +285--+287.
\<horizontal mode material>, 278, 285, 287.
\<horizontal rule>, 281.
horizontal rules, 24, 64, 221--226, 246, +282.
\<horizontal skip>, 285, 290.
Hornschuch, Hieronymus, 483.
Howard, Jane Temple, 21.
|\hphantom|, 178, 211, +360.
*|\hrule|, @24, 64, 85, 221--225, @246, +281--+282, 286, @357, @420, @421.
|\hrulefill| (\hbox to4em{\hrulefill}), @244, @252, +357, @412.
*|\hsize|, @26--27, @60, 102, 188, 237, 251, @257, 274, @340--341, @348,
 @387, @406, @407, 413, 415, @417.
*|\hskip|, 71, 86, 168, 283, +285, 290, @314.
*|\hss|, 71--72, @82--83, 233, 283, 285, 290, 442.
*|\ht|, 120, 271, +388--+389, 417.
Hungarian umlaut (\H{}), \see |\H|.
hyphen, 4, 51, 93, 95, 132, 292.
|hyphen.tex|, 364, +453.
hyphenation, 28, 39, 96, 306, 314, 394--395, 414, 424, +449--+455.
\sub suppressing, 93, 414, 424, 454.
*|\hyphenation|, 277, @419, +452--+453, 455.
\<hyphenation assignment>, +277.
*|\hyphenchar|, 95, 214, +271, 273, +277, 286, 351, 395, @414, 454, 455.
*|\hyphenpenalty|, 96, 101, 272, @348, @451.
\newletter
|\i| ( \i\ ), @52--53, +356.
|I can't go on|, 299.
|\ialign|, +354.
IBM Corporation, vii.
identifiers in programs, 38, 165, @234.
*|\if|, +209, 210--211, 307, 377, @379.
*|\ifcase|, +210, @349, @373, @390, @406.
*|\ifcat|, +209, 210, 307, @377.
*|\ifdim|, +209, @353, @387, @417.
*|\ifeof|, +210, 217.
|\iff| ( $\iff$ ), 163, +361, 438.
*|\iffalse|, +210, 211, @260--261, @348, @385--386.
*|\ifhbox|, +210, @392, @399.
*|\ifhmode|, +209, @363.
*|\ifinner|, +209.
*|\ifmmode|, +209, @215, @240, @353, @356, @360, @423.
*|\ifnum|, @208, +209, @218--219.
*|\ifodd|, @207, +209, @416, @418--419.
*|\iftrue|, +210, 211, @260--261, @348.
|\ifundefined|, 40.
*|\ifvbox|, +210.
*|\ifvmode|, +209.
*|\ifvoid|, +210, @256.
*|\ifx|, +210, 215, 307, @375--377, +384, @418.
ignored characters, 37, 390.
*|\ignorespaces|, +279, @333, @355, @424.
|! Illegal unit|, 295.
illustrations, 115--116.
\sub fitting copy around, 101.
|\Im| ( $\Im$ ), 435.
|\imath| ( $\imath$ ), @136, 435.
*|\immediate|, 226--228, +280, @422, @423.
implicit braces, +269, \see |\bgroup|, |\egroup|.
implicit characters, 269, 277, 309, 351, 376--377.
implicit kerns, 306, +454, 455.
|in| (inch), @24, +57, 270.
|\in| ( $\in$ ), @128, @147, 436.
|\inaccessible|, 297.
inch, 57--58.
incomplete conditionals, 206.
*|\indent|, 86, 94, 101, 263, +282, +286, +291, @355.
indentation or indention, 86, 222, 282.
\sub hanging, 79, 102--103.
indexes, 261--263, 392--394, 423--425, 481.
indices, \see subscripts.
|\inf| ( $\inf$ ), 162, 361.
inferiors, \see subscripts.
infinite badness, 97, 107, 111, 229, 317.
infinite glue, 71--72, 118--119, 256, 313, 332.
infinite loop, 299, 301, 383.
infinite penalty, 97, 111, 254--256, 264, 286, 400.
|\infty| ( $\infty$ ), 9, 318, 435.
inhibiting expansion, 216, 262--263, 377.
|INITEX|, 39, 41, 76, 157, 283, 336, 343--345, 453.
Inner atom, 158, 170, 289, 292, 443, 446.
*|\input|, 7, 9, @25--27, 47, 199, +214, 217, @380, 382--383, @403, @422.
input/output commands, \see |\input|, |\read|, |\write|, |\message|, |\dump|.
input/output streams, 346, \see |\openin|, |\openout|.
input stack, 300, 374.
*|\inputlineno|, 214, 271.
*|\insert|, 95, +122--+125, 259, 280--281, @363, 416, 424, 454.
inserting text online, 31.
insertions, 110, 115--117, 122--125, 256, 335.
*|\insertpenalties|, 111, 114, +123--+125, 214, 254, @256, 271.
|\int| (large $\smallint$), 144, @168--169, @192, +358, 435.
\<integer constant>, +269--+270.
\<integer parameter>, 271, +272--+273, 276.
\<integer variable>, +276.
integral signs, \see |\int|, |\oint|, |\smallint|.
\sub multiple, @169, @180.
inter-column spacing, 237--239, 247, 392.
interacting with \TeX, 31--34, 217--218, 228, 295--299.
\<interaction mode assignment>, +277.
|\interdisplaylinepenalty|, @193, 349, 362.
|\interfootnotelinepenalty|, 349, 363.
interline glue, 78--79, +80, 104, 105, 125, 221, 245, 263, 281--282, 335, 352, 399, 409.
*|\interlinepenalty|, +104, 105, 272, @363, @406, @419.
internal box-and-glue representation, 66, 75, 79, 158--159, 302, 455.
internal character codes, 43--46, 367--370.
\<internal dimen>, +271.
\<internal glue>, +271.
\<internal integer>, +271.
\<internal muglue>, +271.
\<internal quantity>, +279.
Internal Revenue Service, 200, 208--209, 244, 404.
\<internal unit>, +270.
internal vertical mode, 85, 87--89, 222, 278--283.
interrupts, 33, 299, 383.
interword spacing, +74--+76, @356, @393--394, 433.
|Interwoven alignment preambles|, 299.
\<intimate assignment>, +277.
invalid characters, 37, 45.
|\iota| ( $\iota$ ), @325, 434.
Isocrates, 239.
|\it| (use {\it italic\/} type), 13--14, 165, 231--232, @332, +351, @409,
  @414--415, @419, 428.
italic corrections, @14, @64, +287, @306, 441, 455.
italic letters with descenders, 319.
italic type, 13--14, 100, 127, 165, 409, 428, 430.
|\item|, @102--103, 117, @340--342, +355, @416, 419.
|\itemitem|, @102, 342, +355.
iteration, \see |\loop|.
|\itfam|, +351, @414--415.
\t Iur'ev, Serge\u\i\ Petrovich, 53.
\newletter
|\j| ( \j\ ), 52, +356.
Jevons, William Stanley, 5.
|\jmath| ( $\jmath$ ), @136, 435.
*|\jobname|, +213, 214, @336.
Johnson, Samuel, 89.
|\joinrel|, +358.
jokes, vi, 303, 449.
|\jot|, @194, @242, +349, @362.
Joyce, James Augustine, 100.
justification, \see setting glue, line breaking.
Justus, Paul E\period, 455.
\newletter
|\kappa| ( $\kappa$ ), @128, 434.
Kastrup, David Friedrich, 399.
|\ker| ( $\ker$ ), 162, 361.
*|\kern|, 10, 40, @66, 75, 87, 168, @256, @263, +280, @306, @389, @394--395,
  @416, @424, 454--455.
Kernighan, Brian Wilson, 159.
kerns, 4, 66, 75, 95--97, 110, 157, 168, 280, 286, 306, 444, 454--455.
Key Index, 392--394.
keyboards, 3, 5, 43--46, 368--370.
keywords, +61, 71, +268, 337, 370.
Knuth, Donald Ervin, i, ii, vii, 74, 92, 211, 259, 337, 340--341, 412--413,
  424--425.
Knuth, Nancy Jill Carter, iii, 125, 418.
K\"ochel, Ludwig, Ritter von, 409.
\newletter
|l| after |fil|, +271, 337.
{\it\$}, \see pound sterling.
|\l| ( \l\ ), @52--53, +356.
|\L| ( \L\ ), @52--53, +356.
|\lambda| ( $\lambda$ ), @176, 434.
|\Lambda| ( $\Lambda$ ), @323, 434.
Lamport, Leslie B\period, 137.
|\land| ( $\land$ ), 133, +361, 438.
|\langle| ( $\langle$ ), @146--147, 150, 156, +359, 437.
*|\language| (hyphenation method), 273, 346, +455.
large delimiters, 145--150, 442.
large operators, @144--145, 154--155, 358, +435; \also Op atom.
*|\lastbox|, 222, +278, @354, @392, @398, @399.
*|\lastkern|, 214, 271.
*|\lastpenalty|, 214, 271.
*|\lastskip|, 214, @223, 271, 272, @392.
\LaTeX, 137.
|\lbrace| ( $\lbrace$ ), @146--147, +359, 437.
|\lbrack| ( $\lbrack$ ), @146--147, +351, 370, 437.
*|\lccode|, 41, 214, 271, 345, 452--454.
|\lceil| ( $\lceil$ ), @146--147, +359, 437.
|\ldotp|, 358, +359, 438.
|\ldots| ( $\ldots$ ), @73, @172--174, 177, @180--181, @199--201, +359, 438.
|\le| ( $\le$ ), 9, 45, @133, @162, 318, +361, 438.
leaders, 222, 223--226, 228, 280--282, 285, 290, @357, @392--394.
\<leaders>, +281.
*|\leaders|, 95, 110, 223, +224, 225, @357, @392--394.
leading, \see |\baselineskip|, |\vskip|.
|\leavevmode|, 313, @333, @+356, @408, @420.
Lee, Marshall, 17.
*|\left|, @148--150, 155--157, 170, @171, 196, +292, 437.
left brace, @13--14, @19--21, 38, 51, @200--202, +203--+204, @205--206,
  216, +269, 275--276, +283, +286, +291, @330.
\<left brace>, +275.
left bracket, 51, 134, @146--148, 171, @408, @437.
left delimiters, \see openings.
left-hand pages, 252--253, @416.
left parenthesis, 51, 134, @140, @145--150, 345.
left quote, 3--5, 132, 134, @305, 394--395; \also alphabetic constants.
|\leftarrow| ( $\leftarrow$ ), 226, 437.
|\Leftarrow| ( $\Leftarrow$ ), 226, 437.
|\leftarrowfill| (\hbox to4em{\leftarrowfill}), +357.
|\leftharpoondown| ( $\leftharpoondown$ ), 437.
|\leftharpoonup| ( $\leftharpoonup$ ), 437.
*|\lefthyphenmin|, 273, @364, +454, 455.
|\leftline|, 101, @257, @259--260, @326, +353.
|\leftrightarrow| ( $\leftrightarrow$ ), 437.
|\Leftrightarrow| ( $\Leftrightarrow$ ), 437.
*|\leftskip|, 100, 274, @317, @407, @419.
Legendre symbol, 152.
Leontief, Wassily Wassily, 265. % acc to his postcard to me, 25 Feb 1983
|\leq| ( $\leq$ ), 318, +358, 436.
|\leqalignno|, 192, 194, +362.
*|\leqno|, @187, 189, +293, 375--376.
Lesk, Michael Edward, 247.
less than or equal, \see |\le|.
less than sign, 52, 53, @133, 150, 209.
*|\let|, 206--207, 215, +277, 307, 309, @352, @376.
\<let assignment>, +277.
|letterformat.tex|, 403, +406--+408.
letterheads, 407.
letters, 7--8, +37, 41, 45--46, 51, 132, 157, 344, 370.
|\lfloor| ( $\lfloor$ ), @146--147, +359, 437.
|\lg| ( $\lg$ ), 162, 361.
|\lgroup|, 150, @176, +359, 437.
|\lhook|, +358.
Liang, Franklin Mark, 449.
library of macros, 199, 382--384, 425.
lies, vii, 303.
ligatures, 4, 19, 46, 51, 54, 75, 95, 165, 286, 302, 381, 427, 444, 454.
|\lim| ( $\lim$ ), 162, @163, 361.
|\liminf| ( $\liminf$ ), 162, @163, @178, +361.
*|\limits|, 144, 159, +292, @359, 443.
limits above and below operators, 144--145, 149, 179.
|\limsup| ( $\limsup$ ), 162, @163, @178, +361.
Lincoln, Abraham, 11.
|\line|, @72, 77, 101, 224, 232, 252, @255--257, +353, @412.
line breaking, 97--100, 173--174, 392--395, 398--400.
line breaks, avoiding bad, 27--30, 91--94, 173--174, 197.
\sub forcing good, 94, 114.
line rules, \see |\hrule|, |\vrule|.
\<linefeed> (ASCII code 10), 369.
*|\linepenalty|, +98, 272, 314, @316, @348.
lines of input, @24, 46--48, @340.
*|\lineskip|, 78--80, 104, 194, 274, 281, 349, @351--352.
*|\lineskiplimit|, 78--80, 104, 194, 274, 281, 349, @351--352, @362.
list macros, 378--380.
|\listing|, 380, 391.
|\ll| ( $\ll$ ), 431, 436.
Llanfair P\period\ G\period, 451.
|\llap|, 82--83, @189, @340--341, +353, @355, @381, @416--417, @422.
|\lmoustache|, 150, +359, 437.
|\ln| ( $\ln$ ), 162, @186, 361.
|\lnot| ( $\lnot$ ), +361, 438.
local, 19--21, \see grouping.
Locke, John, 303.
|\log| ( $\log$ ), 162, @169--170, 361.
log file, 10, 28, 66, 226, 303, 347.
|\loggingall|, +364.
logical operators, \see |\land|, |\lor|, |\lnot|.
logo, 1, 8, 412, \see |\TeX|.
*|\long|, 205--206, 210, 275, 331, @375, @378, @382.
long formulas, 195--197.
|\longindentation|, 407.
|\longleftarrow| ( $\longleftarrow$ ), +358, 437.
|\Longleftarrow| ( $\Longleftarrow$ ), +358, 437.
|\longleftrightarrow| ( $\longleftrightarrow$ ), +358, 437.
|\Longleftrightarrow| ( $\Longleftrightarrow$ ), +358, 437.
|\longmapsto| ( $\longmapsto$ ), +358, 437.
|\longrightarrow| ( $\longrightarrow$ ), @325, +358, 437.
|\Longrightarrow| ( $\Longrightarrow$ ), +358, 437.
looking ahead, 207, 376--377.
|\loop|, @217--219, +352, @373--374, @387, @417.
loose lines, 97, 99, 302.
*|\looseness|, 103--104, 109, 273, @342, 349.
|\lor| ( $\lor$ ), 133, +361, 438.
low-resolution printer, 59.
*|\lower|, @66, 80, 151, 179, +285, +290.
*|\lowercase|, 41, 215, +279, 307, 345.
lowercase letters, 9, 268, 370, 453.
|\lq| ( \lq\ ), 5, 48, +351, 370, @395.
lxix, 420.
\newletter
|\m@ne|, +347.
|\m@th|, +353.
machine-independence, 58; \also |.dvi|.
Macro, N\ae vius Sertorius, 219.
macro arguments and parameters, 33, 200--205, 249, @263, @362, @363, @375--380,
  @407.
\<macro assignment>, +275.
macro conventions, 121, 346, 364.
macron accent (\={}), \see |\=|, |\bar|.
macros, 199--219, 373--400.
\sub to save typing in math, 136, 199--200, @324.
*|\mag|, +60, 270, 273, 348.
|\magnification|, 17, 59--60, +364, @403--404.
magnified output, 16--17, 59--60, 403, 433.
|\magstep|, 17, 59--60, +349, @403--404.
|\magstephalf|, 17, +349, 403.
mail, 403--408.
main vertical list, 85, 110, +112, 125, 253--254, 281.
|\maintoks|, 262.
|\makefootline|, 255--257, +364.
|\makeheadline|, 255, 257, +364.
|\makelabel|, 403, @404--405, +408.
|manfnt|, 44, 408, 414.
|manmac.tex|, 413--425.
|\manual|, 44, +414, @419--420.
|\mapsto| ( $\mapsto$ ), 128, +358, 437.
|\mapstochar| ( $\mapstochar$ ), 358.
|\margin|, 415, 424.
marginal hacks, 82, 105, 400, 415, 424.
margins, \see |\hoffset|, |\hsize|, |\narrower|.
*|\mark|, 95, 157, 216, 258--263, +280, 417, 454.
marks, 95, 110, 157, 213, +258--+263.
markup commands, \see control sequences.
math accents, @135--137, 141, 164--165, +359, 435, +443.
\<math atom>, +291.
\<math character>, +289.
math character codes, 154--157.
\<math field>, +289.
math fonts, 157, @351, 430--433, 441, 447.
math formulas, how to type, 127--197.
math italic, 164--165, 409, 430, 433.
math lists, 157--159, 441--446.
math mode, 85--89, 127, 157, +289--+293.
\<math mode material>, 287, 289--293.
math spacing table, 170--171.
\<math symbol>, +289.
math symbols, 127--128.
\sub construction of, 151, 178, @358--361.
\sub table of, 434--438.
math-off, 95--97, 287, 447.
math-on, 95, 97, 287, 447.
*|\mathaccent|, 157, 170, +291, @359, +443.
*|\mathbin|, 155, 291, @361.
*|\mathchar|, 155, 289.
*|\mathchardef|, 155, 199, 214, 215, 271, 277, 289, 336, @358, @394.
\<mathchardef token>, +271, 289.
*|\mathchoice|, 151, 157, 292.
*|\mathclose|, 155, 291, @322, @359.
*|\mathcode|, 134, +154--+155, 214, 271, 289, 319, 326, @344.
mathematical expressions, 127--197.
Mathematical Reviews, 106, 392--394.
mathematical style, 166--167, 182--183.
|\mathhexbox|, +356.
*|\mathinner|, 155, 171, 199, 291, @359.
*|\mathop|, 155, 178, 291, @324--325, @361.
*|\mathopen|, 155, 291, @322, @359.
*|\mathord|, 88--89, 155, 291.
|\mathpalette|, 151, @+360.
*|\mathpunct|, 155, 291.
*|\mathrel|, 155, 291, @359--361.
|\mathstrut|, 131, 178, +360.
*|\mathsurround|, 97, @162, 274, 305, 314, 323, @353, 447.
matrices, 176--178, 181; \also alignments.
|\matrix|, @176--178, 182, @325, +361.
|\max| ( $\max$ ), 162--163, @170--171, 361.
*|\maxdeadcycles|, +255, 273, @348.
*|\maxdepth|, @112--114, 123--125, @255, @262--263, 274, @348, 400, @415.
|\maxdimen|, 58, 188, @262--263, +347, 348.
maximum legal dimension, 58.
maximum legal integer, 118.
*|\meaning|, +213--+215, 336, @382.
measure, \see |\hsize|.
|\medbreak|, 111, 113, +353, @355, @419, @422.
*|\medmuskip| (medium math space), 167--168, 274, @349, 446.
|\medskip| (medium extra vertical space), 70, 79, 102, 109, 111, +352,
  @410--412.
|\medskipamount|, +349, @352--353, @355, @407.
|\medtype|, 408--411.
membership, \see |\in|, |\ni|, |\notin|.
memory space, 100, 300--301, 342, 345, 347, 383, 384, 394, 399.
*|\message|, 216, @217--218, 227--228, +279, @308, 328, @343--344, @355, @418.
{\manual opqrstuq}, 420, 427, 442, 483. % METAFONT (eight-point type)
metric units, 57--60.
|\mid| ( $\mid$ ), @174, 436.
|\midinsert|, 116, @340--341, +363.
migration, 105, 117, 259, 280--281, 282, 286, 287.
|\min| ( $\min$ ), 162--163, @170--171, 361.
|minus|, 71, 271, 337.
minus sign, 4, 51, @127, 132, 226, 268.
|! Missing| something, 296--297, 337.
|\mit| (math italic family), 164, +351, 430, @434.
*|\mkern|, 168, +280, +442.
|mm| (millimeter), +57, 270.
mod, @164, 322, +361.
|\models| ( $\models$ ), +358, 436.
modes, 46, 85--89, 175, 267--293.
|modes.tex|, 88--89.
money, 54, 140, @208--209.
*|\month|, 273, 349, @406.
Morris, William, 107.
moustaches, 150.
mouth, \see anatomy of \TeX.
*|\moveleft|, 80--81, +282, 287.
*|\moveright|, 80--81, @221, +282.
Moxon, Joseph, 287.
Mozart, Johann Chrysostom Wolfgang\rlap{ Gott-}\break % kludge to save a line
 lieb (= Theophilus = Amadeus), 409.
|\mp| ( $\mp$ ), @133, 436.
*|\mskip|, 168, +290, 442.
|mu| (math unit), 168, 270, 442.
|\mu| ( $\mu$ ), @162, 434.
\<mu unit>, +270.
\<mudimen>, +270.
\<muglue>, 118, 167--168, +271.
\<muglue parameter>, 271, +274, 276.
\<muglue variable>, +276.
Muir, Cockburn, 283.
Muirhead, James, 21.
multicolumn format, 257, 396--397, 417.
multiple integrals, @169, @180.
*|\multiply|, @118--119, @218, +276, 349, @391, @398.
|\multispan|, 243, 246--247, @334, +354.
Munster, 55.
\<mushrink>, +271.
music, 408--412.
*|\muskip| registers, 118, 168, +271, +276.
*|\muskipdef|, 119, 215, +277.
\<muskipdef token>, +271.
\<mustretch>, +271.
\newletter
|\nabla| ( $\nabla$ ), 435.
names, 73, 92.
|\narrower|, 100, @340--341, +355.
National Science Foundation, vii.
|\natural| ( $\natural$ ), 435.
natural width, 69.
Naur, Peter, 268.
|\ne| ( $\ne$ ), 9, 45, @133, 318, +361, 438.
|\nearrow| ( $\nearrow$ ), 437.
|\neg| ( $\neg$ ), 435.
negated relations, 436--437.
negative dimensions, 66, 222.
|\negthinspace|, 332, +352.
|\neq| ( $\neq$ ), 318, +358, 438.
nesting (i\period e\period, groups inside groups), 20--21, 210, 385.
new symbols, 151, 178, @358--361.
|\newbox|, 121, 346, +347, @353, @394, @417.
|\newcount|, 121, @218, 346, +347, @349, @418.
|\newdimen|, 121, 346, +347, @349, @415.
|\newfam|, 121, 157, 346, +347, @351.
|\newhelp|, 346, +347.
|\newif|, 211, @218, +348, @354, @375, @416, @423.
|\newinsert|, 121, 122, 346, +347, @363, @415.
|\newlanguage|, 346, +347.
*|\newlinechar|, +228, 273, @348.
|\newmuskip|, 121, 346, +347.
|\newread|, 121, 216, 346, +347.
|\newskip|, 121, 346, +347, @349, @394, @414.
|\newtoks|, 121, 212, @262, 346, +347, @400.
|\newwrite|, 121, 227, 346, +347, @422--423.
|\next|, +352.
|\ni| ( $\ni$ ), 436.
|\ninepoint|, 15, +414, @419.
|\ninerm|, 15, 413.
|\ninesl|, 15, 413.
|No room|, 347.
*|\noalign|, 176, 191, @193, +237, @246, 249, +282, +285, 286.
*|\noboundary|, 283, +286, 290.
|\nobreak| (inhibit line or page break), 97, 109, @174, @193, 335,
  +353, @394, @407.
*|\noexpand|, +209, +213, 215, 216, @377, @424.
*|\noindent|, 86, 188, @262--263, +283, 286, 291, @340--341, @355, @419.
|\nointerlineskip|, 79--80, @255, 331, +352, @389.
*|\nolimits|, 144, 159, +292, @358, @361.
\<non-macro assignment>, +275.
nonaligned leaders, 224--226.
|\nonfrenchspacing|, 74, +351.
*|\nonscript|, 179, +290, 442, 446.
*|\nonstopmode|, +32, +277, 299.
|\nopagenumbers|, 251--252, +362, @406, @409.
norm symbol, \see absolute value, vertical line.
\<normal dimen>, +270.
\<normal integer>, +269.
\<normal mudimen>, +270.
|\normalbaselines|, @325, 349, +351, @414--415.
|\normalbaselineskip|, +349, @414--415.
|\normalbottom|, +363.
|\normallineskip|, +349, @351.
|\normallineskiplimit|, +349, @351, @362.
Norwegian characters, 45--46, 52--53, 370.
|\not| ( $\not$ ), @133, +358, @436--437.
|\notin| ( $\notin$ ), +360, 437.
$n$th, 323.
|\nu| ( $\nu$ ), @128, @163, 434.
nucleus, 158--159, 289--292, 441--446.
\<null> (ASCII code 0), 37, 39, 48, 343, 369.
|\null|, 311, @312, @316, @332, @335, +351, @354, @360--@362, @419.
null control sequence, 46, 308.
null delimiter, 149--150, 152, 156, 345, @360, @362.
null set, \see |\emptyset|.
*|\nulldelimiterspace|, 150, 274, @348, +442.
*|\nullfont|, 14, 153, 271, 433.
\<number>, 44, 118, +269--+270, 272, 309, 349.
*|\number|, 40--41, +213, 214, @252, @406, @424.
number sign, \see hash mark.
numbered footnotes, 121, 125.
numerals, \see digits, roman numerals.
numerator, 141, 152, 179, 444--445.
\<numeric variable>, +276.
numerical tables, 240--241.
|\nwarrow| ( $\nwarrow$ ), 437.
\newletter
|O| versus |0|, 132.
|\o| ( \o\ ), @52, +356.
|\O| ( \O\ ), @52, +356.
|\oalign|, +356.
obelisk {\sl or\/} obelus, 53.
|\obeylines|, @94, @249, @262, 342, +352, @380--382, @407, @419.
|\obeyspaces|, 254, 308, 342, +352, @380--381, @394, @421.
oblique, \see slanted.
\<octal constant>, +269.
\<octal digit>, +269.
octal notation, +43, @44, 155, 420.
odd-numbered pages, 252--253, @416.
|\odot| ( $\odot$ ), 436.
|\oe| ( \oe\ ), 52, +356.
|\OE| ( \OE\ ), 52, @53, +356.
Office of Naval Research, vii.
|\offinterlineskip|, @245--247, 312, +352, @416.
|\oint| (large \ooalign{$\smallint$\cr\hidewidth$\circ$\hidewidth}), +358, 435.
|\oldstyle|, +351.
oldstyle numerals, 430, 435.
|\omega| ( $\omega$ ), @323, +358, 434.
|\Omega| ( $\Omega$ ), +358, 434.
|\omicron|, 434.
|\ominus| ( $\ominus$ ), 436.
*|\omit|, 240, 243--244, @246--247, 282.
one half, 141, 186.
\<one optional space>, +269--+270.
online interaction, \see interacting with \TeX.
|\ooalign|, +356.
Op atom, 158--159, 170--171, 289, 442--444.
Open atom, 158, 170--171, 289, 442--444, 446.
open intervals, 171.
openface, \see blackboard bold.
*|\openin|, +216--217, 280.
openings, 134, 147, 154--155, 359, +437; \also Open atom.
*|\openout|, 226--228, 254, +280, @422, @423.
|\openup|, 194, 237, 242, +362.
|\oplus| ( $\oplus$ ), 9, 154, 436.
optimization of macros, 342, 345, 348.
\<optional assignments>, +286.
\<optional {\tt by}>, +276.
\<optional sign>, +57.
\<optional signs>, +269.
\<optional spaces>, +268, 269--271.
\<optional {\tt true}>, +270.
*|\or|, 210, +213, @406.
|\Orb| ( \Orb\ ), +356.
Ord atom, 158, 170--171, 289, 358, 442--446.
ordinary symbols, 132, \see Ord atom.
Ore, \O ystein, 53.
organists, 239.
organs, 38--39, 46, 85, +267, 373, +456.
orphans, \see widow words.
Osbourne, Lloyd, 67.
|\oslash| ( $\oslash$ ), 436.
|\other|, 421.
other character, 37.
|\otimes| ( $\otimes$ ), 9, 436.
*|\outer|, +206, 210, 275, @354, @357, @418--419, @422.
*|\output|, 125, +253, @254--257, 275, @364, 370, @417.
output routines, 21, 112, 251--264, 417.
\sub when invoked, 122, 125, 281.
*|\outputpenalty|, +125, @254--255, 273, 349, 400, @417.
*|\over|, @139--141, @148, 152, +292, 437, 444--445.
Over atom, 158, 289, 443.
|\overbrace|, @176, 225, +359.
overfull boxes, 27--30, 94, 229, 238, 302--303, 307, 400.
\sub avoiding, 107.
*|\overfullrule|, 274, 307, @348.
overlaps, 82--83, 386.
|\overleftarrow|, +359.
*|\overline|, @130--131, @136, 141, 170, 291, 443.
overloading, 54, 243.
|\overrightarrow|, 226, +359.
*|\overwithdelims|, @152, +292, 444--445.
|\owns| ( $\owns$ ), +361, 438.
\newletter
|\P| ( \P\ ), 53, 117, +356, 438--439.
|\p@|, +347, 348.
|\p@renwd|, +361.
page breaks, avoiding bad, 109--111, 189, 193.
\sub forcing good, 109--111; \also |\eject|.
page builder, 110--114, 122--125, 281.
\sub when exercised, 122, 280--283, 286--287.
page format, modifying, 251--253.
page make-up, 109, \see output routines.
page numbers, 21, 23, 119, 207, 251--253.
|\pagebody|, 255--257, +364.
|\pagecontents|, 256, +364.
*|\pagedepth|, +114, 123, 214, 271, 272.
*|\pagefilllstretch|, +114, 214, 271, 272.
*|\pagefillstretch|, +114, 214, 271, 272.
*|\pagefilstretch|, +114, 214, 271, 272.
*|\pagegoal|, +114, 123, 214, 271, 272.
|\pageinsert|, 115, +363.
|\pageno|, @252, @256, @340, +362, @406.
*|\pageshrink|, +114, 123, 214, 271, 272.
*|\pagestretch|, +114, 214, 271, 272.
*|\pagetotal|, +114, 123, 214, 271, 272.
*|\par|, +47, 86--87, 100, 135, 202, 249, 262, +283, +286, @340, 351, @380--381.
\sub forbidden in arguments, 205, 207.
paragraph, implied beginning of, 85--86, +283.
\sub implied end of, @24, 86, +286.
\sub last line of, 99--100.
\sub shape parameters reset, 103.
paragraph sign, 53.
|\parallel| ( $\parallel$ ), 436.
\<parameter text>, 203, 275.
parameters, \see macro arguments and parameters.
parameters, numeric, 119, 342; \also |\fontdimen|.
\sub default values, 348--349.
\sub table, 272--274.
parentheses, 51, 129, 134, @140, @145--150, 345, 437.
\sub roman, in italic text, 409--411.
*|\parfillskip|, 100, 188, 274, +286, 307, @315, 332, @348, @394, @419.
*|\parindent|, 86, 100, 101--102, 105, @262, 274, 282, 286, 291, @342, @348,
  @355, @394, @406, @415.
*|\parshape|, 101--103, 214, 271, 272, 277, 283, @315, 349, 374.
*|\parskip|, 79, 104--105, 262, 274, 282, @342, @348, @355, @406, @417.
|\partial| ( $\partial$ ), @147, 435.
Pascal, Blaise, 101--102.
|\path|, 390--391.
*|\patterns|, 277, +453, 455.
patterns for hyphenation, 449--453.
*|\pausing|, 273, +303.
Paxton, William Hamilton, 425.
|pc| (pica), +57, 270, @415.
penalties, 95--100, 110--114, 189.
\sub infinite, 97, 111, 286, 400.
\sub negatively infinite, 97, 111--112, 114, 254--256, 264, 400.
*|\penalty|, 79, 97, 110--111, 174, +280, @353.
|\penalty-'10000000000|, 264.
percent sign, @26, 38, 39, 43, 48, 51, 113, 124, @249, @337, @340.
period, 51, @72--73, 133--134, 149, @161, 345, @394--395, 430.
\sub space after, @73--75, +76.
periodic preambles, 241--242, @244, @246, @361, @412.
Perlis, Alan Jay, 365.
|\perp| ( $\perp$ ), +358, 436.
|\phantom|, 131, 178, 211, +360, @412.
|\phi| ( $\phi$ ), @128, @148, 434.
|\Phi| ( $\Phi$ ), @323, 434.
Philips, Peter, 239.
philosophers, 100, 239.
\<physical unit>, +270.
|\pi| ( $\pi$ ), 9, @87, @137, @148--149, 434.
|\Pi| ( $\Pi$ ), 9, @324, 434, 435.
piano, 17.
pica, 57, 413, \see |pc|.
pictures, 228--229.
pieces of symbols, 145, 432, 442.
pilcrow, \see |\P|.
|plain.tex|, 342--364.
plain \TeX\ format, 10--11, +343--+364.
\sub summary, 339--342.
|\plainoutput|, 255, +364.
Plass, Michael Frederick, 340--341.
Plato, 1, 239.
|plus|, 71, 271, 337.
\<plus or minus>, +268.
plus sign, 51, 132, 268.
|\pm| ( $\pm$ ), @133, +358, 436.
|\pmatrix|, 176, @323, +362.
|\pmod| (notation like (mod~$p$)), @164, 322, +361.
|\pnt|, 395.
pocket-size, 59.
poem, 94.
points (printers' units), 15, 57--58.
points with arbitrary coordinates, 389.
Polish characters, 52--53.
P\'olya, Gy\"orgy (= George), 7.
poor man's bold, 386.
*|\postdisplaypenalty|, 189--190, 272.
poultry, 236--237.
pound sign, \see hash mark.
pound sterling, 54, @339, 428.
powers of ten, 293.
powers of two, 293.
|\Pr| ( $\Pr$ ), 162, @323, 361.
Pr\ae torius [Schultheiss], Michael, 239.
preambles, 206, 235--249.
|\prec| ( $\prec$ ), 436.
|\preceq| ( $\preceq$ ), 436.
*|\predisplaypenalty|, 189--190, 272, @348.
*|\predisplaysize|, 188, 190, 274, 349.
\<prefix>, +275.
|\preloaded|, +350, 413.
preloaded formats, 25--26, 39, 283, 344.
Presume, Livingstone Irving, 74.
*|\pretolerance|, +96, 107, 272, @317, @348, @364, @394, @451.
pretty-printed programs, 165, 234.
*|\prevdepth|, 79--80, 89, 271, 272, 281, +282.
|prevdepth ignored|, 88--89.
*|\prevgraf|, 103, 188, 190, 214, 271.
|\prime| ( $\prime$ ), @130, 155, 357, 435.
prime numbers, 148--149, 218.
primitive, 9--11, 267, 342, 457.
private control sequences, 344, 364, 414.
|\proclaim|, 202--203, 206, @340--341, +355.
Procter, William Alexander, 38.
|\prod| (large $\Pi$), @180--181, 435.
programming with \TeX, 217--219, 387--388.
programs, for computers, 38, 165, @234.
\sub for music, 408--412.
proofreading, 59, 303.
proper names, 73, 92.
|\propto| ( $\propto$ ), 436.
prototype row, 238, 302--303.
|\ps|, 403, @404, +407.
pseudo parameters, 119, +349.
|\psi| ( $\psi$ ), @325, 434.
|\Psi| ( $\Psi$ ), 434.
psychologically bad breaks, @91--93.
|pt| (printer's point), @24, +57--+58, 268--270.
Punct atom, 158, 170--171, 289, 442--443, 446.
punctuation, 14, 51, 72--76, 321, 394--395.
\sub in formulas, 134, 154--155, 161, 358--359, 438; \also Punct atom.
\newletter
|\qquad|, @166, @185, +352.
quad, 60, 166--168, 177, 433.
|\quad|, @94, 166--167, @185, @232--233, +352.
quad left, \see flush left.
quad middle, \see |\break|.
quad right, \see flush right.
quarter circles, 389--391.
question mark, 51, 73, 161.
Quick, Jonathan Horatio, 298.
quotation marks, @3--5, @24, 394--395.
quotations, 100, 418--419, 425.
quotes within quotes, 5.
\newletter
$\rm I\!R$, \see blackboard bold.
Rad atom, 158, 289, 443.
*|\radical|, 157--159, 291, 443.
ragged bottom margins, 111, 253, 256, 398.
ragged left margins, 392--394.
ragged right margins, 29--30, 101, 261--262, 355--356, 392--394.
|\raggedbottom|, 111, 253, +363, @406.
|\raggedcenter|, 107.
|\raggedright|, 29--30, 76, 101, 107, @115, @262, +356, @396, @407.
*|\raise|, @66--67, 80, 151, 179, @193, +285, +290, @408.
|\rangle| ( $\rangle$ ), @146--147, 150, +359, 437.
Raper, Matthew, 61.
|\rbrace| ( $\rbrace$ ), @146--147, +359, 437.
|\rbrack| ( $\rbrack$ ), @146--147, +351, 370, 437.
|\rceil| ( $\rceil$ ), @146--147, +359, 437.
|\Re| ( $\Re$ ), 435.
*|\read|, 215, 217--218, 276, 346, 400.
recent contributions, 112, 125, 281.
recovery from errors, 30--34, 295--303.
recursion, 219, 268, 301, 379, 391, 397.
\sub infinite, 299, 301, 383.
reduction, 16.
reference marks, 116--117.
reference point of a box, 63--64, 77, 80--82, 388--389.
\<registerdef>, +277.
registers, 117--122, 212, 214, 346--348.
registration marks, 416.
Reid, Brian Keith, 371, 404--406.
Rel atom, 158, 170--171, 289, 442--444, 446--447.
\<relation>, +209.
relations, @133--134, 147, 154--155, 358, +436, +437; \also Rel atom.
*|\relax|, 23, 25, 71, 240, 276, +279, @307, @353.
|\relbar| ( $\relbar$ ), +358.
|\Relbar| ( $\Relbar$ ), +358.
*|\relpenalty|, 101, 174, 272, 322, @348, +446.
|\removelastskip|, +353.
|\repeat|, @217--219, +352.
repeating commands, \see |\loop|.
repeating templates, \see periodic preambles.
replacement text, @200--@204, 212, 280, 300, 329.
reserved characters, 37--38, 51--52, 134.
reserved words, +61, 71, +268, 337, 370.
restricted horizontal mode, 85, 87--89, +285--+287.
r\'esum\'e, 253.
\<return> (ASCII code 13), 23, 39, 43, 45, 46, 48, 249, 331, 343, 345, 369, 380.
|\|\<return>, 8, @305, +351.
reverse apostrophe, 3--5, 51, 132, 134, 391, 394--395;
  \also alphabetic constants.
reverse slash, \see backslash.
Reviewer, Ann Arbor, 106.
|\rfloor| ( $\rfloor$ ), @146--147, +359, 437.
|\rgroup|, 150, @176, +359, 437.
|\rho| ( $\rho$ ), 128, @325, 434.
|\rhook|, +358.
*|\right|, @148--150, 155--157, @170--@171, 196, +292, 437.
right brace, @13--14, @19--21, 38, 51, @200--202, +203--+204, @205--206,
  +269, 275--276, +279, 301, @330.
\<right brace>, +275.
right bracket, 51, 134, @146--147, 171, 345, @408, @437.
right delimiters, \see closings.
right-hand pages, 252--253, @416.
right justification, 71.
right parenthesis, 51, 134, @140, @145--150, 345.
right quote, 3--5, 51, @130, 155, 201, @305, 324, 394--395; \also octal.
|\rightarrow| ( $\rightarrow$ ), 226, 437.
|\Rightarrow| ( $\Rightarrow$ ), 226, 437.
|\rightarrowfill| (\hbox to4em{\rightarrowfill}), 226, +357.
|\rightharpoondown| ( $\rightharpoondown$ ), 437.
|\rightharpoonup| ( $\rightharpoonup$ ), 437.
*|\righthyphenmin|, 273, @364, +454, 455.
|\rightleftharpoons| ( $\rightleftharpoons$ ), +361, 437.
|\rightline|, 101, @317, @340--341, +353.
*|\rightskip|, 100--101, 274, @317, @356, 393, @421.
|\rlap|, 82--83, @189, @247, @319, +353, @389, @416.
|\rm| (use roman type), 13--15, 154, @163, @320, +351, @364, @409, @414--415,
  @419, 427.
|\rmoustache|, 150, +359.
roman letters in math, 162--164.
roman numerals, 40--41, 252, 256.
\sub uppercase, 374.
roman parentheses in italic text, 409--411.
roman type, 13--17, 51--55, 127, 162--165, 427.
*|\romannumeral|, 40--41, 213, 214, @252.
Root, Waverley Lewis, 55.
|\root|, @130--131, @179, @325, +360.
rounding, 58, 119.
|\rq| ( \rq\ ), 5, 48, +351, 370, @395.
rule boxes, 24, 64, 221--225, 281--282, 285, 291.
\<rule dimension>, +282.
\<rule specification>, +282.
rule thickness, 143, 179, 221, 447.
ruled tables, 245--248, 392.
rulers, 58.
runaways, 205--206, 297.
running headlines, 253, 258--260, 416.
running the program, 23--35.
Russian characters, 370.
\newletter
|\S| ( \S\ ), 53, 117, +356, 438--439.
Sally, Baby, 72--76.
sample line for tabbing, 232--234.
sample pages for book design, 412--413.
|save size|, 300--301, 374, 399.
save stack buildup, 301, 346.
|\sb|, 135, +357, 370.
|\sc| (use {\sc SMALL CAPS} type), 414--415.
|scaled|, 16--17, 60, 277, @350, 433.
scaled points, 57--58, 270.
Scandinavian letters, 45--46, 52--53.
Schickele, Prof\period~Peter, 410--411, 481.
scratch control sequence, 352.
scratch registers, 122, 346.
Scribe, 371, 404.
script letters, 164, 434.
script size, 140, 153, 442.
script style, 140--142, 441--447.
*|\scriptfont|, 153, 168, 213, 271, 321, 351, @414--415, 441--442.
scriptscript size, 140, 145, 153, 442.
scriptscript style, 140--142, 441--447.
*|\scriptscriptfont|, 153, 168, 213, 271, 351, @414--415, 441--442.
*|\scriptscriptstyle|, 141--142, @179, 292.
*|\scriptspace|, 274, 348, 445--446.
*|\scriptstyle|, 141--142, @145, @179, 292.
*|\scrollmode|, +32, +277.
|\searrow| ( $\searrow$ ), 437.
|\sec| ( $\sec$ ), 162, 361.
section number sign, \see |\S|.
Selden, John, 11.
selection, \see |\cases|.
semicolon, 51, @134, 161.
sentences, 72--76.
Sesame Street, 61.
set macro, 379.
set notation, @147, @174--175.
*|\setbox|, 66--67, 77, 81, +120, 276, 279, 286, 291, @386--392.
*|\setlanguage|, 287, +455.
|\setminus| ( $\setminus$ ), @436.
|\settabs|, 231--234, +354, 355.
setting the glue, 70, 77, 81, 388.
|\setupverbatim|, 380--381, 391.
|\sevenrm|, 15, 153, +350, 351, @414--415.
*|\sfcode|, +76, 214, 271, 286, @321, @345, @351.
shadow boxes, 66.
Shakespeare, William, 17, 55, 337.
\<shape assignment>, +277.
\<shape dimensions>, +277.
|\sharp| ( $\sharp$ ), @409, 435.
sharp S ( \ss\ ), \see |\ss|.
sharp sign, \see hash mark.
Shaw, George Bernard, 107, 229.
Sheridan, Richard Brinsley Butler, 265.
shifted output, \see |\hoffset|, |\voffset|.
shilling sign, \see slash.
*|\shipout|, 227, 253--254, 279, 300, 302.
shortest paths, 99.
\<shorthand definition>, +277.
*|\show|, @10, 215, 279, 299.
*|\showbox|, @66--67, 121, 234, 279.
*|\showboxbreadth|, 273, +302, 303, @348.
*|\showboxdepth|, 79, 273, +302, 303, @348.
|\showhyphens|, +364, @452.
*|\showlists|, @88--89, 95, 112, 125, @158--159, 279, 293.
*|\showthe|, 121, 215, 279.
shriek, \see exclamation point.
\<shrink>, +271.
shrinkability, 69--71, @75, 409.
side conditions, @166--167, @185--186.
|\sigma| ( $\sigma$ ), @195--196, 434.
|\Sigma| ( $\Sigma$ ), @165, 434, 435.
sigma signs, \see |\sum|.
|\signed|, 106.
|\sim| ( $\sim$ ), @133, 436.
|\simeq| ( $\simeq$ ), @133, 436.
\<simple assignment>, +276.
|\sin| ( $\sin$ ), 162, 361.
|\sinh| ( $\sinh$ ), 162, 361.
sinkage, 413.
size switching, 15, 408, 414--415.
sizes of type for mathematics, 140, 153.
|\skew|, @136, +359.
*|\skewchar|, 214, 271, 273, 277, @351, @414, 430, 431, +443.
*|\skip| registers, 118--122, +271, +276, @346--347, @349, @352, @363, @394.
*|\skipdef|, 119, 215, +277, @346--347.
\<skipdef token>, +271.
skipping space, \see glue.
|\sl| (use {\sl slanted\/} type), 13--15, 165, +351, @409, @414--415, @419.
slant of a font, 375, 433.
slanted type, 13, 63--64, 100.
slash, 51, @132, 146--147, @320, 430.
|\slash| (/ with break allowed), @93, +353.
slashed form of fractions, 139--140.
Slavic h\'a\v cek accent, \see |\v|.
|\slfam|, +351, @414--415.
slides, 59.
slurs, \see |\smile|, |\frown|.
small caps, 414.
|\smallbreak|, 111, +353, 421.
|\smallint| ( $\smallint$ ), +358, 435.
|\smallskip|, 70, 78, @100, 109, 111, @181, @340--341, +352, @355, @410--412.
|\smallskipamount|, +349, @352--353.
|\smalltype|, 408--411.
|\smash|, 131, 178, @327, +360.
|\smile| ( $\smile$ ), 436.
Smith, John, 447.
solidus, \see slash.
|sp| (scaled point), +57, 118--119, 270, 398, 400.
|\sp| (superscript), 135, +357, 370.
\<space> (ASCII code 32), 39, 43, 46, 343.
|\|\<space>, \see |\|\] (near the beginning).
|\space|, 254, 272, @306, +351, 380, @406.
space after a constant, 208, 272.
space factor, 76, 285--287, 306, 380, 395.
\<space token>, 268, +269, 282, 285, 290, @376--377.
*|\spacefactor|, 76, 271, 272, 285, @363, 433.
spaces, 3, 5, 8--9, 19, 37, 40, 47--48, 127, 204, 232, 272, 297, 299, 319, 336.
\sub as active characters, 254, 380--381.
*|\spaceskip|, 76, 274, @317, @356, @416, 429.
spacing, \see glue.
\sub in formulas, 162, 167--171.
|\spadesuit| ( $\spadesuit$ ), +358, 435.
*|\span|, 215, 238, +243, @244, +245, 248, 249, 282, @330, 385.
Spanish ligatures, 51, 381, 427.
spanned columns in tables, 243--245.
spanned rows in tables, 249.
*|\special|, 216, 226, +228--+229, 280.
special characters, 37--38, 43--46, 51--52, 134, 367--371.
\<special dimen>, +271, 277.
\<special integer>, +271, 277.
special symbols for math, 128, 434--438.
Spivak, Michael David, 137.
|% split|, 124.
split insertion penalty, 124.
*|\splitbotmark|, 213, +259, 280.
*|\splitfirstmark|, 213, +259, 280.
*|\splitmaxdepth|, 124, 274, 281, @348, @363, @417.
*|\splittopskip|, 124, 274, 281, @348, @363, @397, @417.
|spread|, 77, 222, 238, 278.
springs, 70.
|\sqcap| ( $\sqcap$ ), @133, 436.
|\sqcup| ( $\sqcup$ ), @133, 436.
|\sqrt|, @130--131, @141, @145, 157, @169--170, +360, +443.
|\sqsubseteq| ( $\sqsubseteq$ ), 436.
|\sqsupseteq| ( $\sqsupseteq$ ), 436.
|\square| ( $\vcenter{\hrule
      \hbox{\vrule height\sqht \kern\sqht \vrule}
      \hrule}$ ), 151.
square brackets, \see brackets.
square bullet ( \bull\ ), +420.
square root, \see |\sqrt|.
squeeze routine, 188.
squiggle accent (\~{}), \see |\~|, |\tilde|, |\widetilde|.
squint print, 59.
|\ss| ( \ss\ ), 52, +356.
stack positions, 300--301, 374, 399.
stacked fractions, \see |\over|.
|\star| ( $\star$ ), 436; {\sl cf\period}~asterisk.
states, 46--48.
sterling, 54, @339, 428.
Stevenson, Robert Louis Balfour, 67.
Stirling numbers, \see |\brace|, |\brack|.
stomach, \see anatomy of \TeX.
stopping \TeX, \see |\end|.
|story.tex|, 24, 26, 30--31.
Stout, Rex Todhunter, 401.
\<stretch>, +271.
stretchability, 69--71, @75, 409.
*|\string|, 40--41, +213--+214, 215, @348, @377.
|\strut|, 82, @142, 178, 240, @246--247, @316, @329, @333, +353, @396, @400,
  @421.
|\strutbox|, @316, +353, @396, @414--415.
struts, +82, 125, 131, 142, 155, 178, 245--247, 255, 329, 416, 422, 423.
style change items, 157, 442.
style-independent documents, 194, 203.
styles of math formatting, 140--141, 441--447.
styles of math writing, 166--167, 182--183.
subformulas, 129, 171, 173.
\<subscript>, +291.
subscripts, 15, @128--131, @133, 150, 158--159, @163, 179, +289--+291, 343.
|\subset| ( $\subset$ ), @133, 436.
|\subseteq| ( $\subseteq$ ), @133, 436.
|\succ| ( $\succ$ ), 436.
|\succeq| ( $\succeq$ ), 436.
|\sum| (large $\Sigma$), @139, @144--145, @148--149, +358, 432--433, 435.
|\sum|$'$, 181, @324--325.
summary of plain \TeX, 339--342.
summation, \see |\sum|.
|\sup| ( $\sup$ ), 162, +361.
|\supereject|, 116, 254, 256, 257, +353, @407.
superiors, \see superscripts.
\<superscript>, +291.
superscripts, @128--131, @133, 150, 158--159, 179, +289--+291, 343.
suppressed-L, 52--53.
|\supset| ( $\supset$ ), 436.
|\supseteq| ( $\supseteq$ ), 436.
|\surd| ( $\surd$ ), +358, 435.
surd signs, \see |\radical|, |\sqrt|.
Swanson, Ellen Esther, 197.
|\swarrow| ( $\swarrow$ ), 437.
Swedish characters, 52--53, 370.
Sweelinck, Jan Pieterszoon, 239.
\'Swierczkowski, Stanis\l aw S\l awomir, 53.
Swift, Jonathan, 5.
symbol fonts, 157, @351, 431, 433, 441, 447.
symbolic box format, 66, 75, 79, 158--159, 302, 455.
symbols in math, table, 434--438.
syntax rules, 268--269.
System Development Foundation, vii.
Szeg\H o, G\'abor, 7.
\newletter
|\t| (tie-after accent), 52--53, +356.
\<tab> (ASCII code 9), 8, 45, 343, 369, 391.
|\|\<tab>, 8, 351.
|\tabalign|, +354, 355.
tabbing, 231--234, @339, +354.
\sub compared to |\halign|, 235.
tables, \see alignments, tabbing.
tables of contents, 120, 226.
tables of \TeX\ trivia:
\sub accents (non-math), 52, 339.
\sub atomic types, 158.
\sub category codes, 37.
\sub character codes, 367--369.
\sub default values, 343--345, 348--349.
\sub |\fontdimen| parameters, 433, 447.
\sub keywords, 61.
\sub math accents, 135.
\sub math classes, 154.
\sub math spacing, 170--171.
\sub math symbols, 434--438.
\sub parameters, 272--275.
\sub units of measure, 57.
|\tabs|, 234, +354.
*|\tabskip|, 215, +237--+239, @244, @247, 274, 282, 285, @354.
tabskip glue, +237--+239, 245, 302--303, 392.
Tacitus, Publius Cornelius, 219.
tags, \see equation numbers.
tail recursion, 219.
|\tan| ( $\tan$ ), 162, 361.
|\tanh| ( $\tanh$ ), 162, +361.
|\tau| ( $\tau$ ), 1, 434.
Tchaikovsky, \see Cha\u\i kovski\u\i.
telephone numbers, 408.
templates, 206, 235--236, 240--243.
|\tenex|, +350, @351, @361, @414--415.
|\tenpoint|, 15, +414, @418.
|\tenrm|, 15, 27--28, 45, 153, 252, @335, +350, @351, @414.
|\tensl|, 15, +350, @351, @414.
tensor notation, @130, @169.
|\tensy|, @179, +350, @351, @414.
terminal, input from, 217--218.
\sub output to, 217--218, 226--228, 279--280.
TEX, 1.
\TeX, bad puns on, 1, 11, 63, 153, 161, 225, 229, 262, 305, @340--341, +373.
\sub pronunciation of, 1.
|TeX|, 1.
|\TeX| ( \TeX\ ), 8--10, 19, 66--67, 204, 225, @340--341, +356, +418, +419.
|.tex|, 25, 217, 226.
|TeX capacity exceeded|, 300--301.
\TeX\ Grandmasters, 253.
\TeX\ Users Group, vii, 408, 483.
\TeX78, vii.
|texput|, 23.
text between aligned displays, 193.
text italic, 13--14, 157, 165, 332, 430.
text size, 140, 153, 442.
text style, 140--142, 441--447.
*|\textfont|, 153, 168, 188, 213, 271, +351, @414--415, 441--442.
|\textindent|, 117, +355.
*|\textstyle|, 141--142, +292, @326.
|.tfm| (font metric files), 400, 433, 442.
*|\the|, +214--+215, @216, @373, @375, @422.
theorems, 111, 202--203, @340--341, +355.
|\theta| ( $\theta$ ), 128, @162, @325, 434.
|\Theta| ( $\Theta$ ), 434.
*|\thickmuskip| (thick math space), 167--168, 274, @349, 446.
thin spaces, 161, 167--173, 305, 409.
*|\thinmuskip| (thin math space), 167--168, 274, @349, 446.
|\thinspace|, @5, 10, 305, @311, +352, 409.
|This can't happen|, 299.
Thor, Arthur Uther, 24, 117, 200, 253.
three-column output, 257, 396--398.
three dots, \see ellipses.
tie-after accent, \see |\t|.
ties, 25, 73--74, +91--+93, 104, @161, 167, @173, 353, @404.
tight lines, 97, 99, 302.
tilde, 38, 51, @91--92, \see ties.
|\tilde| (math accent: $\tilde x$), 135, 164.
tilde accent (\~{}), \see |\~|, |\tilde|, |\widetilde|.
*|\time|, 273, 349.
|\times| ( $\times$ ), @133, 436.
Times Tribune, Peninsula, 455.
Titelouze, Jehan, 239.
|to|, 77, 217, 222, 238, +276, +278.
|\to|, @134, +361, 438.
|\today|, @406.
token list parameters and registers, 212, 215, 275, 373.
token lists, as displayed by \TeX, 228, 329, 382.
\<token parameter>, +275.
\<token variable>, 212, +276.
tokens, 38--41, 46--48, 203--207.
*|\toks|, 212, 215, 262, 276.
*|\toksdef|, 212, 215, 277, @347, @378.
\<toksdef token>, +271.
*|\tolerance|, @29--30, 91, 94, +96, 107, 272, @317, @333, @342,
  @348, @364, @451.
|\top| ( $\top$ ), 435.
|\topglue|, @340, +352.
|\topins|, 256, +363, @364.
|\topinsert|, 115--116, 251, +363.
*|\topmark|, 213, +258, 259--260, 280.
*|\topskip|, +113--+114, 124, 256, 274, @348.
Trabb Pardo, Luis Isidoro, 92.
|\tracingall|, 121, 303, +364.
*|\tracingcommands|, 88--89, +212, 273, 299.
*|\tracinglostchars|, 273, +301, @348, @400.
*|\tracingmacros|, +205, +212, 273, @329.
*|\tracingonline|, 121, 212, 273, +303.
*|\tracingoutput|, +254, 273, @301--302.
*|\tracingpages|, @112--114, 124, 273, +303.
*|\tracingparagraphs|, @98--99, 273, +303.
*|\tracingrestores|, 273, +301, +303.
*|\tracingstats|, 273, +300, +303, @383.
transcript, \see log file.
transparencies, 59.
|\triangle| ( $\triangle$ ), 435.
|\triangleleft| ( $\triangleleft$ ), 436.
|\triangleright| ( $\triangleright$ ), 436.
triangular quotation, 101--102.
tricky macros, 41, 261--263, 354, 360, 361, 373--400, 421.
|true|, 59--60, +270, @407--408.
truth, vi, 267.
Tschichold, Jan, 83.
|\tt| (use {\tt typewriter} type), 13, 53, 113, 165, @340--341, +351,
  @380--382, @414--415, +429.
|\ttfam|, +351, @414--415.
|\ttglue|, 414--415, 421.
|\ttraggedright|, +356.
{\sl\backup TUGboat}, vii, 425, 483.
turtle commands, 390--391.
Twain, Mark (= Clemens, Samuel Langhorne), 83, 451.
two-column format, @257, 386--388, @417.
two-line displays, 196.
type size switching, 15, 408, 414--415.
typefaces, 13, 17, 427.
typewriter type, 13, 53, 165, 356, 420--421, 429.
\newletter
|\u| (breve accent), 52--53, +356.
*|\uccode|, 41, 214, 271, 345, @348, @377, @394.
*|\uchyph|, 273, @348, +454.
umlaut accent (\"{}), \see |\"|, |\ddot|.
unbreakable spaces, \see ties.
|\uncatcodespecials|, 380.
|\undefined|, 350, 384.
undelimited parameters, 203--204, 376--377.
Under atom, 158, 289, 443.
|\underbar|, @244, @323, +353.
|\underbrace|, @176, 225--226, +359.
underfull box, 29, 94, 238, 302--303, 397, 400, 417, 452.
*|\underline|, @130--131, 141, 291, +443.
underlined text, 178; \also |\underbar|.
underscore (the character `|_|'), 38, 51, @128--130, 134, @165.
*|\unhbox|, 120, 283, +285, 293, @354, @356, @399.
*|\unhcopy|, 120, 283, +285, 293, @353.
\<unit of measure>, 268, +270.
units of measure, 57--61, 270.
\sub in formulas, 169.
\sub table, 57.
*|\unkern|, +280.
unmatched left brace, 216; \also runaways.
*|\unpenalty|, +280.
unset box, 240, 302--303.
\<unsigned dimen>, +270.
\<unsigned mudimen>, +270.
\<unsigned number>, +269.
*|\unskip|, 222--223, +280, 286, @313, @392, @418--419.
*|\unvbox|, 120, 254, +282, 286, @354, @361, @363, @364, @392, @399, @417.
*|\unvcopy|, 120, +282, 286, @361.
|\up|, 408.
|\uparrow| ( $\uparrow$ ), @146--147, @182, +359, 437.
|\Uparrow| ( $\Uparrow$ ), @146--147, +359, 437.
|\upbracefill| (\hbox to4em{\upbracefill}), 225--226, +357.
|\updownarrow| ( $\updownarrow$ ), @146--147, +359, 437.
|\Updownarrow| ( $\Updownarrow$ ), @146--147, +359, 437.
|\uplus| ( $\uplus$ ), 436.
*|\uppercase|, 41, 215, 217, +279, 307, 345, @348, @374, @377, @394.
uppercase letters, 9, 64, 268, 370.
\sub Greek, 127, 164, 434.
\sub roman numerals, 374.
|\upsilon| ( $\upsilon$ ), 434, 483.
|\Upsilon| ( $\Upsilon$ ), 434.
Ursa Major, 389.
User, Ben Lee, 74, 80, 100, 140, 164, 186, 187, 193, 221, 296, 410--411.
\newletter
|\v| (check accent), 52, +356.
*|\vadjust|, 95, 105, 109, 110, 117, 259, +281, @317, 393, 454.
*|\valign|, 249, 283, +285--+286, 302, @335, @397.
Vall\'ee Poussin, Charles Louis Xavier Joseph de la, 92.
van der Waerden, Bartel Leendert, 92.
van Leunen, Mary-Claire, 425.
|\varepsilon| ( $\varepsilon$ ), 128, 434.
\<variable assignment>, +276.
variable family, 154, 289, @358--359.
variables in formulas, 132, 358, 434--435; \also Ord atom.
|\varphi| ( $\varphi$ ), 128, @147, 434.
|\varpi| ( $\varpi$ ), 434.
|\varrho| ( $\varrho$ ), 128, 434.
|\varsigma| ( $\varsigma$ ), 434, 483.
|\vartheta| ( $\vartheta$ ), 128, 434.
*|\vbadness|, 272, @348, @397, @417.
vbox (box with vertical list inside), 65.
*|\vbox|, 65, 80--82, 103, 151, 193, 222, 278, 388--389.
Vcent atom, 158, 290, 443.
*|\vcenter|, 150--151, 159, 170, 193, 222, 242, +290, @326, @361, +443.
|\vdash| ( $\vdash$ ), 436.
|\vdots| ( $\vdots$ ), @177, +359.
|\vec| (math accent: $\vec x\,$), 135; \also |\overrightarrow|.
vectors, 177, 199--201; \also |\vec|.
|\vee| ( $\vee$ ), @133, 436.
verbatim copying, 422--423.
verbatim listing, 48, 380--382, 391, 420--421.
|\vert| ( $\vert$ ), @146--147, 150, +359, 438; {\sl cf\period}~|\mid|.
|\Vert| ( $\Vert$ ), 117, @146--147, 150, 186, +359.
\<vertical command>, +286.
vertical line (the character `\|'), 52, 53, 132, @146--147, @171, @174, 438;
  \also vertical rules.
vertical lists, 64, +110.
vertical mode, 85--89, +267--+283.
\<vertical mode material>, 278, 280--282, 290.
\<vertical rule>, +281.
vertical rules, 64, 151, 221--226, +285, +291, 392.
\<vertical skip>, +281.
very loose lines, 97, 99.
*|\vfil|, 71, +72, 111, 256, 281, 286, 417.
*|\vfill|, 24, 25, 71, +72, 256--257, 281, 286.
|\vfilll|, 72.
*|\vfilneg|, +72, 111, 281, 286.
|\vfootnote|, 117, +363.
*|\vfuzz|, 274, @348.
|\vglue|, +352, @408.
vinculum, \see |\overline|.
virgule, \see slash, 51.
visible space, +3, +420, 429.
*|\voffset|, +251, 252--253, 274, @342, @406.
void, 120, 210.
|\voidb@x|, +347, 348.
Voltaire, de (= Arouet, \hfil\penalty0\hfilneg Fran\c cois Marie), 447.
|\vphantom|, @178--179, 211, @321, +360.
*|\vrule|, 64, 86, 151, 221--222, 224, @245--247, +281--+282, 283, @357, @392,
  @420.
*|\vsize|, 113--114, 251, 253, 255, 274, @340--341, @348, 400, @406, 413, @415,
  @417.
*|\vskip|, @24, 71, 85, 191, +281, 286.
*|\vsplit|, +124, 222, 259, 278, @397, @417.
*|\vss|, 71, +72, @255, 281, 286.
*|\vtop|, 81--82, 151, 222, 278, @333.
\newletter
Walter's worksheet, 244.
Walton, Izaak, 67.
Watson, John Hamish, M\period D\period, 401.
*|\wd|, 120, 271, +388--+389, @391, @417.
|\wedge| ( $\wedge$ ), @133, 436.
Weierstrass $p$, \see |\wp|.
weird error, 298.
Welsh, 239, 451.
Westing, Arthur Herbert, 246.
whatsits, 95, 110, 157, 226--229, 455.
|\widehat| (math accent: $\widehat x\,$), @136, +359, 433.
|\widetilde| (math accent: $\widetilde x\,$), @136, +359.
widow lines, 104, 272, 398.
widow words, 104.
*|\widowpenalty|, +104, 105, 113, 272, @348.
|width|, 221, 224, +282, 337.
width of a box, 63--67, 77, 80--82, 225.
wiggle, \see |\sim|.
Winograd, Terry Allen, 425.
Wischmeyer Ehlert, Clara Louise, 248.
|withdelims|, 152, 156.
|\wlog|, +347.
Wolfe [Holmes], Nero, 401.
|\wp| ( $\wp$ ), 435.
|\wr| ( $\wr$ ), 436.
wreath product, \see |\wr|.
*|\write|, 215, 216, 226--228, 254, +280, 346, @377, @422, @424.
\newletter
$\times$, \see |\times|, |\*|.
x-height, 54, 60, 433, 443, 445--447.
*|\xdef|, @215--216, 275, @373, @418, @424.
Xenophon, 239.
|\xi| ( $\xi$ ), 434, 447.
|\Xi| ( $\Xi$ ), 434.
*|\xleaders|, +224.
*|\xspaceskip|, 76, 274, @317, 356, 429, 433.
|\xyzzy|, 379.
\newletter
*|\year|, 41, 273, 349, @406.
Yellin, Frank Nathan, 400.
\newletter
|\z@|, +347, 348.
|\z@skip|, +347, 348.
|\zeta| ( $\zeta$ ), 434.
\newletter
\frac1/2, 67, 332.
1/2, in unslashed form, 141, 186.
\<4-bit number>, +271.
\<8-bit number>, +271, 276--278.
\<15-bit number>, +271, 277, 289, 291.
\<27-bit number>, +271, 289, 291.
\enddoublecolumns
\endchapter

Important works such as histories, biographies,
scientific and technical text-books, etc., should contain indexes.
Indeed, such works are scarcely to be considered complete without indexes.
\smallskip
An index is almost invariably placed at the end of a volume
and is set in smaller type than the text-matter.
Its subjects should be thoroughly alphabetized.
\smallskip
The compiling of an index is interesting work, though
some authors are apt to find it tedious and delegate the work to others.
The proofreader who undertakes it will find that it is splendid mental exercise
and brings out his latent editorial capability.
\author ALBERT H. ^{HIGHTON}, {\sl Practical Proofreading\/} (1926)

\bigskip

Important references are given in boldface.
Italicized numbers indicate fleeting references,
whereas numbers in parentheses refer to
mere implications or unwarranted extrapolations.
Asterisks are used to identify particularly distasteful passages.
\author PROF.~PETER ^{SCHICKELE}, {\sl The Definitive Biography of %
   P. D. Q. ^{Bach}\/} (1976)

\eject
\beginchapter Appendix J. Joining the\\\TeX\ Community

This appendix is about grouping of another kind: \TeX\ users from around the
world have banded together to form the \TeX\ Users Group (TUG), in order
to exchange information about common problems and solutions.

A newsletter/journal called {\sl TUGboat\/} has been published
since 1980, featuring articles about all aspects of \TeX\ and \MF\null.
^^{METAFONT}
TUG has a network of ``site coordinators'' who serve as focal points of
communication for people with the same computer configurations.
Occasional short courses are given in order
to provide concentrated training in special topics; videotapes of
these courses are available for rental.
Meetings of the entire TUG membership are held at least once a year.
You can buy \TeX\ T-shirts at these meetings.

Information about membership in TUG and subscription to {\sl TUGboat\/}
is available from
\begintt
{\obeylines
\TeX\ Users Group
email: {\tt TUG@tug.org}
internet: {\tt http://www.tug.org/}
}
\endtt
Don't delay, subscribe today! That address again is

\smallskip
{\obeylines
\TeX\ Users Group
email: {\tt TUG\char`\@ tug.org}
internet: {\tt http://www.tug.org/}
}

\endchapter

[The printer] should refuse to employ wandering men,
foreigners who, after having committed some grievous error,
can easily disappear and return to their own country.
\author HIERONYMUS ^{HORNSCHUCH}, %
  {\eightrm'}$O\mkern-1mu\rho\mkern1mu\vartheta o\mkern1mu %
  \tau\upsilon\pi o\gamma\mkern-1mu %
  \rho\alpha\phi\acute\iota\alpha\varsigma$ (1608)

^^|\upsilon| ^^|\varsigma|
\bigskip

An author writing an article for publication in TUGboat
is encouraged to create it on a computer file and submit it on magnetic tape.
\author BARBARA ^{BEETON}, {\sl {\tt\bslash title} How to Prepare a File\/%
   {\tt\bslash cr}\phantom{\rm\enskip(1981)}\break %
   For Publication in TUGboat\/{\tt\bslash cr}} (1981) % TUGboat 2,1 p53

\eject
\end